JP4348577B2 - Motion capture device using myoelectric potential information and control method thereof, as well as electrical stimulation device, force-tactile sensation display device using the same, and control method thereof - Google Patents

Description

【０１６４】
尚、表１において「○」は横欄に示す各ＤＢが縦欄に示すＤＢを包含することを意味し、「−」はそのような包含関係がないことを意味している。例えば、（Ａ）の骨に関するＤＢには、骨の重量や重量分布、骨の破断係数に関するＤＢが含まれる。また、（Ｂ）のＤＢについては関節の重量が主として骨部の重量から構成されるために靭帯についての重量や重量分布を無視しているが、これらについてもモデルに組み込むことでより詳細なモデル化を図ることができることは勿論である。
【０１６５】
そして、形状に関するＤＢについては、基準人体（上記基準体型モデル等。）の骨格や脂肪等の形状モデルに基づき前記した方法を用いて体型形や性別データ等から生成される。
【０１６６】
図２９及び図３０は、上記ステップＳＳ１及びＳＳ２について、具体例の要部を示すフローチャート図である。
【０１６７】
先ず、図２９のステップＳ１において予め規定されている下記のデータ項目について、対象者の数値を入力（手動又は自動入力）したり、選択値の場合にはそれらのいずれかを指定する。
【０１６８】
・身長（単位：ｍｍ）
・体重（単位：Ｋｇ）
・体型形（痩せ型、闘士型、肥満型等のタイプとその度合。）
・性別（男・女）
・年齢（単位：才）。
【０１６９】
例えば、入力値のデータ構造について下表２に示す例が挙げられる。
【０１７０】
【表２】

【０１７１】
つまり、この場合には、身長や体重等のように連続的な数値を格納する領域を要するものや、体型形等のように、タイプを示す値（上記体型形座標空間の座標軸θの値に対応する）とその度合示す数値（上記体型形座標空間の座標軸ｒの値に対応する）とを組み合わせた構造、あるいは、性別のように「０」又は「１」の１ビットデータで簡単に表現できるもの等が挙げられる。
【０１７２】
尚、身体構造モデルの生成にあたっては、上記のような比較的少数の入力パラメーターに基づいてモデルを生成するモード（以下、「ノーマルモード」という。）と、ノーマルモードで生成したモデルに変更を加えるためのモード（以下、「特殊モード」という。）とが存在するので、ステップＳ２でのモード判断処理において、先ずは、ノーマルモードを選択したものとして、ステップＳ３乃至Ｓ８での処理について説明する。
【０１７３】
ステップＳ３では、上記（ａ）の重量ＤＢを参照して重量設定処理を行う。即ち、対象者の体型形データに基づいて骨、筋肉、脂肪、頭部、臓器等についての重量比率を設定するとともに、筋肉や脂肪の重量についてはさらに胸部、腹部、上肢、下肢に区分して設定する。また、性別の如何によって筋肉脂肪のつき方が異なるので、その相違を考慮して重量比率の設定を行う。尚、設定後における総重量と上記した体重の入力値との差がほぼゼロとなるように重量の割り当てを行う必要があることは勿論である。
【０１７４】
次ステップＳ４では、上記（ｂ）の重量分布ＤＢを参照して、対象者の身長や体型形のデータに基づいて骨格や脂肪の形状について設定を行う。そして、骨の重量分布（重心や比重等）により骨と頭部の重量配分を設定するとともに、肉質の重量分布により筋肉や脂肪の重量配分を設定する。尚、ここで、「肉質の重量分布」について、上肢、下肢の場合と胸部の場合とに分けて説明すると、前者の場合には、骨の重量分布についての重量点を中心とした仮想の円軌道（これは立位姿勢の人体を正面から見た場合の形状であり、正確には球形状をしている。）を複数設定して、各軌道に対して重量を等間隔でもって均等に配分する。また、後者の場合には骨の重量分布における胸郭の重量点から外方へ一定の間隔をおいた距離に仮想の軌道を複数設定して、各軌道に対して重量を等間隔でもって均等に配分する。そして、性別の入力データが女性である場合には、胸部脂肪の重量分布について追加の設定を行う。
【０１７５】
内臓の重量分布については、頭内臓部、第１頸骨間関節と恥骨間を結ぶ内臓重量線上において等間隔で設定し、該重量分布は椎骨の運動によって変化する。また、腹部脂肪の重量分布については、内臓重量線上に沿って等間隔に設定される重量点を中心とする仮想の楕円形軌道（これは立位姿勢の人体を正面から見た場合の形状であり、正確には楕円体形状をしている。）を複数設定し、各軌道に対して重量を等間隔でもって均等に配分する。尚、この他、仮想の楕円形軌道内に、骨、筋肉、内蔵、脂肪の領域を設けて領域毎のＤＢを作っておき、各領域に等分布の重量をそれぞれ設定する方法がある。
【０１７６】
上記によって内臓や脂肪の分布が明らかとなるので、対偶の進入禁止範囲（対偶が内臓等を突き抜けて体内に進入するのを禁ずるための範囲）を設定することができる。
【０１７７】
ステップＳ５では、上記（ｃ）の破壊係数ＤＢから骨や靭帯、筋肉等についての破壊係数データを読み込む。これは骨や靭帯等の破損を招くような無理な力が加わる姿勢、つまり、健康体においてあり得ない人体の状態を人体構造モデルにおいて回避するために必要とされる。
【０１７８】
ステップＳ６では、上記（ｄ）の関節自由度ＤＢや（ｃ）の破壊係数ＤＢを参照して、体型形や性別のデータから関節の駆動範囲や自由度（破壊時の自由度を含む。）を設定する。尚、これは関節の構造からは健康体として許されない動きやありえない動きを排除するためである。
【０１７９】
ステップＳ７では、上記（ｅ）接続関係ＤＢを参照して各種の接続関係についてのデータを読み込んだ後、次ステップＳ８では筋運動についての設定を行う。
【０１８０】
筋運動については上記（ｆ）の筋運動ＤＢを参照して、入力データ（身長、体重、体型形、性別、年齢等）に基づいて駆動される対偶の重量を考慮して関節の駆動角度と作用筋の収縮率について設定する。
【０１８１】
以上のステップＳ３乃至Ｓ８によってノーマルモードでの身体構造モデル及びこれをデータベース化した身体構造ＤＢが作成される。尚、上記（ｇ）の神経配置ＤＢについては、身体構造ＤＢに合わせて基準人体（基準体型モデル等）の神経配置から単独の加工処理によって作成される。
【０１８２】
図３０のステップＳ９では特殊モードに進むか否かを判断し、特殊モードを選択した場合にはステップＳ１０に進み、選択しなければステップＳ１７に進む。
【０１８３】
上記ステップＳ２又はステップＳ９において特殊モードを選択した場合にはステップＳ１０に進み、比率変換モードを選択するか否かを判断する。尚、「比率変換モード」とは上記したノーマルモードにおいて生成したデータに対して腕、胴、脚部等の大きさを入力してデータ比較を行い、長さや重量の再設定を行うモードである。同ステップで比率変換モードを選択した場合にはステップＳ１１に進み、選択しない場合にはステップＳ１７に進む。
【０１８４】
ステップＳ１１では比率変換モードを更に２つにモード、つまり、「長さ設定モード」と「比率設定モード」とに分け、両者のうちのいずれかを選択する。そして、「長さ設定モード」を選択した場合にはステップＳ１２に進んで、腕、胴、脚部等の長さ（単位：ｍｍ）をそれぞれ入力する。また、「比率設定モード」を選択した場合にはステップＳ１３に進んで腕、胴、脚部等の基準長（例えば、身長等）に対する比率をそれぞれ入力する。尚、これらステップＳ１２、Ｓ１３での入力データは、例えば、上記（ｅ）の接続関係ＤＢにおける筋肉の自由長についての補正に用いられる。
【０１８５】
続くステップＳ１４では、重量モードについて２つのモード、つまり、「合わせ込みモード」と「加減調整モード」とに分け、両者のうちのいずれかを選択する。そして、「合わせ込みモード」を選択した場合にはステップＳ１５に進み、生成データから得られる重量を体重の入力データに合わせる処理を行ってから図２９のステップＳ２に戻る。また、「加減調整モード」を設定した場合にはステップＳ１６に進んで生成データから得られる重量について増減量を付与する処理を行った後ステップＳ２に戻る。これらのステップＳ１５、Ｓ１６により、上記ステップＳ３やＳ４での重量設定や重量分布の補正をさらに行うことができるようになる。
【０１８６】
ステップＳ１７では、「関節重量モード」を選択するか否かを判断し、選択する場合にはステップＳ１８に進み、選択しなければステップＳ１９に進む。尚、「関節重量モード」とは、例えば、対象が人体ではなく人体を模倣した２足歩行型ロボットにおいて、関節駆動用のモータ等の重量を登録する場合等に用いられ、関節毎に１点の加重値を加減算して、全体のバランスが保たれるように対偶の重量分布を設定するためのモードである。尚、ステップＳ１８での処理後はステップＳ１９に進む。
【０１８７】
ステップＳ１９では、「筋肉重量モード」を選択するか否かを判断し、選択する場合にはステップＳ２０に進み、ここで各対偶に付随する筋や脂肪の重量毎に所望の加重値を加減算することで筋肉等の重量バランスを保つように設定を行う。
【０１８８】
尚、「筋肉重量モード」を選択しなければ身体構造モデルの生成を終了するが、これまでの工程において必要なデータは体格や性別等に関する比較的少数のデータだけである。
【０１８９】
次に上記したステップＳＳ３、ＳＳ４について、図３１に示す要部のフローチャート図に従って説明する。尚、本工程では、上記（Ｇ）身体運動に関するＤＢについて下記に示すＤＢが作成されるが、その内容は後述する処理の説明から明らかとなる。
【０１９０】
（Ｇ１）運動ワイヤーフレームＤＢ
（Ｇ２）運動重心位置ＤＢ。
【０１９１】
先ず、図３１のステップＳ１において重心位置検出用マーカーを付設した対象者に対して立位姿勢を維持してもらった後、次ステップＳ２では３次元データの取得法を用いて対象者の３次元データを得る。尚、ここで、「３次元データ」とは、平面上に貼り付けられた２次元画像と区別されるデータであり、３次元形状を構成する曲面及びこれに貼り付けられた画像データを意味する。
【０１９２】
３次元データの取得法としては、被写体の形状や画像を含むデータを得ることのできる方法であれば如何なる方法を用いても構わないが、例えば、前記した光の干渉縞を用いた方法を挙げることができる。
【０１９３】
次ステップＳ３では、前ステップによって得られたモデルに対してその頭部から足先に向かって鉛直方向に沿ってスライス処理（つまり、断層断面を形成する。）を行うことで身体の各構成部分や対偶を認識する。例えば、図３２に概略的に示すモデルｍｄｌに関して矢印Ｒで示す方向がスライス方向であり、点Ｐｓが頭頂部を示し、領域「Ａ１」が頭部の認識に関する領域、領域「Ａ２」が頸椎の認識に関する領域、領域「Ａ３」が肩部の認識に関する領域、領域「Ａ４」が胸部及び腹部の認識に関する領域、領域「Ａ５」が脚部の認識に関する領域を代表的に示している。尚、図３２において右側に概略的に示した断層図は領域Ａ１の下側境界面でスライスした領域の数が１個、領域Ａ３の下側境界面でスライスした領域の数が３個、脚部を途中でスライスした領域の数が２個であることをそれぞれ示している。
【０１９４】
また、対偶等の認識については、対象者の身体構造モデルと３次元データから得られるモデルとを比較・対照することで行う。例えば、被写体（対象者）に取り付けた重心位置検出用マーカーに対応する重心と頭部の中心点とを結ぶ軸を身体構造モデルの背骨軸であると認識したり、上肢の両肩関節の位置については背骨軸に関して左右対称であって最も突起した箇所として認識する等、人体の身体的特徴に基づいて求めることができる。
【０１９５】
図３１のステップＳ４では対象者に対して膝の屈伸運動を行ってもらい、上記ステップＳ２及びＳ３で説明したのと同様の方法によって３次元データの取得及び膝関節の認識に関する処理を行う。つまり、立位姿勢での３次元データだけでは膝関節の特定を正確に行うことができないためである。
【０１９６】
そして、ステップＳ５では対象者に対してその前腕が地面に平行になるように腕を動かしてもらって上記ステップＳ２、Ｓ３で説明したのと同様の方法によって３次元データの取得及び肘関節の認識に関する処理を行う。
【０１９７】
次ステップＳ６では上記ステップＳ２乃至Ｓ５において得られた基礎データについての補正（誤差の補正等）を行った後、次ステップＳ７に進み、データの確定、つまり、対象者に係る人体形状データを決定する。
【０１９８】
そして、ステップＳ８に進み、対象者の形状と上記ステップで得た人体形状データとの間に顕著な相違が生じないように形状について両者間の合わせ込み処理を行い、当該人体形状と上記身体構造モデルとの間に生じる差異を極力低減する。
【０１９９】
ステップＳ９では、対象者に所定の運動をしてもらい、上記ステップＳ２で説明したのと同様の方法を用いて運動状態についての３次元データを取得する。
【０２００】
そして、ステップＳ１０では３次元データにおける頭部と重心位置検出用マーカーの位置に基づいて胴体を認識するとともに、対象者の身体運動に関する情報（対偶の状態等）を表象する枠体モデルとしてワイヤーフレームモデルを生成する。そして、該ワイヤーフレームモデルをデータベース化することにより上記の運動ワイヤーフレームＤＢを作成する。
【０２０１】
ステップＳ１１では、ワイヤーフレームモデル及び前記ステップで得られた人体形状に基づいて運動中心である重心位置及びその変化に関するデータだけを抽出して上記運動重心位置ＤＢを作成する。
【０２０２】
しかして、これまでに経た処理によって身体の構造や運動に関する数値モデルを得ることができ、これら対して筋肉や脂肪の形状についてのポリゴンデータ（多角形近似あるいは多面体近似によるデータ）を付加することによって人体に関するポリゴンモデルやこれをデータベース化したもの（以下、「人体ポリゴンＤＢ」という。）を得ることができる。
【０２０３】
図３３は人体ポリゴンＤＢについて、上記した身体構造ＤＢや運動ワイヤーフレームＤＢ、運動重心位置ＤＢの他、下記に示すＤＢとの依存関係の一例を示すものであり、図中の矢印「→」は、「Ｘ→Ｙ」と記した場合にデータベース「Ｙ」がデータベース「Ｘ」に基づいて生成されることを意味し、また、両矢印はリンクを意味している。
【０２０４】
（ｈ）筋肉ポリゴンＤＢ
（ｉ）脂肪ポリゴンＤＢ
（ｊ）特殊脂肪ポリゴンＤＢ
（ｋ）皮膚老化ポリゴンＤＢ。
【０２０５】
上記（ｈ）乃至（ｊ）のＤＢについて簡単に説明すると、先ず、（ｈ）筋肉ポリゴンＤＢは筋肉の収縮に対応した筋肉形状のポリゴンデータ集であり、（ｉ）脂肪ポリゴンＤＢは、筋肉の収縮と重心の運動に対応した脂肪形状のポリゴンデータ集である。また、（ｊ）特殊脂肪ポリゴンＤＢとは、筋肉の収縮には直接関係しないが重心の運動に主として関与する脂肪（内臓脂肪や胸部脂肪等。）の形状についてのポリゴンデータ集である。
【０２０６】
（ｋ）皮膚老化ポリゴンＤＢは、皮膚に対して人為的に老化させる処理を施す際に必要な皺の量や寄り方についてのポリゴンデータ集（あるいは２次元画像データ）であり、上記した（Ｆ）皮膚に関するＤＢを構成するものである。
【０２０７】
尚、図３３に示す「筋収縮ポリゴンＤＢ」は、筋肉ポリゴンＤＢ、身体構造ＤＢ、運動ワイヤーフレームＤＢ、運動重心位置ＤＢから生成されるデータベースであり、人体の運動に伴う筋肉の収縮状態を表現するために必要とされ、概ね下記の手順に沿って生成される。
【０２０８】
（１）身体構造ＤＢの生成
（２）運動ワイヤーフレームＤＢ及び運動重心位置ＤＢの生成
（３）筋肉ポリゴンＤＢと（１）や（２）のＤＢとのリンク。
【０２０９】
また、「脂肪収縮ポリゴンＤＢ」は、筋肉ポリゴンＤＢ、身体構造ＤＢ、運動ワイヤーフレームＤＢ、運動重心位置ＤＢを参照しながら脂肪ポリゴンＤＢ及び特殊脂肪ポリゴンＤＢから生成されるデータベースであり、筋肉の収縮や重心運動に伴う脂肪の位置や厚さ等の変化を表現するために必要とされる。
【０２１０】
図３４は、人体ポリゴンＤＢの生成について処理例の要部を示すフローチャート図であり、筋収縮ポリゴンＤＢの生成に関する処理と、脂肪収縮ポリゴンＤＢに関する処理とを並列的に示している。
【０２１１】
先ず、筋収縮ポリゴンＤＢの生成に関する処理については、ステップＳ１において、筋肉ポリゴンＤＢ、身体構造ＤＢ、運動ワイヤーフレームＤＢ、運動重心位置ＤＢを用意した後、ステップＳ２では、人体の運動と各筋肉の収縮率の変化についての関係を得る。
【０２１２】
そして、ステップＳ３では全ての筋肉に関してその収縮に対応した筋肉形状を示すポリゴンデータ集、つまり、筋収縮ポリゴンＤＢを作成する。その際には、身体構造ＤＢのうち特に人体形状に関するデータ及び筋肉ポリゴンＤＢのデータを参照する。
【０２１３】
次ステップＳ４では、運動ワイヤーフレームＤＢに対して筋肉部のポリゴンデータを付加した後、ステップＳ５に進む。
【０２１４】
他方、脂肪、特殊脂肪のポリゴンＤＢに関する処理については、ステップＳＴ１において両ＤＢを用意した後、次ステップＳＴ２で人体の運動と各脂肪の収縮率の変化についての関係を得る。
【０２１５】
それからステップＳＴ３において全ての脂肪に関して筋肉の収縮や重心運動に対応した脂肪形状を示すポリゴンデータ集、つまり、脂肪収縮ポリゴンＤＢを作成する。その際に身体構造ＤＢのうち特に人体形状に関するデータ及び筋肉ポリゴンＤＢのデータを参照する必要がある。
【０２１６】
次ステップＳＴ４では、運動ワイヤーフレームＤＢに対して脂肪部のポリゴンデータを付加した後、ステップＳ５に進む。
【０２１７】
ステップＳ５では、筋肉や脂肪を加味した運動ワイヤーフレームモデルについて、画像データの１フレーム毎に筋肉や脂肪のポリゴンデータを加工することにより筋肉の躍動や脂肪の揺れ等を表現する。尚、この加工にはポリゴンの表面におけるドットやパッチに対するデータ処理として行われ、例えば、ドットに関する結線処理やベヂエ化処理、あるいはこれらの処理を、一定表面の圧縮後に行う等の処理が含まれる。
【０２１８】
次ステップＳ６では上記皮膚老化ポリゴンＤＢと運動ワイヤーフレームＤＢとのリンクを行う（図３３では両方向の矢印で示している。）。これによって、例えば、ある関節の周囲における皮膚の皺を関節角度や関節からの距離に応じて変化させることができる。
【０２１９】
こうして人体ポリゴンＤＢが作成されるが、筋肉への電気刺激にあたっては、例えば、下記の（イ）乃至（ハ）に示す手順を踏む。
【０２２０】
（イ）駆動の対象となる関節を特定する
（ロ）（イ）の関節を動作させるのに必要な筋肉（例えば、対をなす筋肉として伸筋及び屈筋等）又は刺激付与の対象となる筋肉を選定する
（ハ）（ロ）で選定した筋肉において、皮膚表面に近い表層の筋に対しては効果器を通して低周波による刺激（あるいは後述するファントムセンセーションによる刺激）を与え、内層の筋に対しては効果器を通して干渉波による刺激を与える。
【０２２１】
つまり、刺激を与える筋の位置に応じて電気刺激の態様（干渉波電気刺激又は低周波電気刺激）を使い分けることで、それぞれの筋に対して効率良く刺激を与え、違和感のない運動感覚を装置使用者に呈示することができる。
【０２２２】
尚、上記手順（イ）では駆動の対象となる関節を特定したが、効果器によっては当該効果器に係る対象関節が既に特定されている場合があるので、そのときには、直ちに筋肉を選定することができる。
【０２２３】
また、（ロ）において、関節の動作に必要な筋として伸筋及び屈筋を選定した場合には伸筋と屈筋とを同じ程度に収縮させるのではなく、関節の屈曲時には屈筋をより多く収縮させ（例えば、屈筋対伸筋の収縮比率を２０：１程度の比とする。）、また、これとは逆に関節の伸展時には伸筋の方をより多く収縮させる。そして、筋肉の選定においては、関節とこれを駆動する筋肉の伸縮とを関係付けるデータベースを予め構築しておき、その後、上記（イ）の過程で、対象となる関節を特定したときに、当該関節を駆動するのに必要な筋肉を上記データベースに基づいて選定することが好ましい。尚、「関節とこれを駆動する筋肉の伸縮とを関係付けるデータベース」とは、例えば、前記した（Ｂ）関節や靭帯に関するＤＢと（Ｃ）筋肉に関するＤＢとを関連付けることによって生成することができ、（ａ）乃至（ｆ）に示すＤＢのうち、関節や対偶と、これを補助する筋肉についてのデータ項目（例えば、関節の名称や位置、関節の駆動に要する筋肉の名称や形状データ、対偶を支えている始点の関節名称等）を含んでいる。このデータベースを用いると、例えば、肘関節を対象としたときに、その動作に応じて関与する筋肉の情報を得ることができる。
【０２２４】
例えば、肘関節の駆動をとりあげて説明すると、以下のようになる。
【０２２５】
人体における肘関節及び前腕の運動に関与する筋のうち、上腕の筋肉には、上腕二頭筋、上腕筋、上腕三頭筋、肘筋が挙げられ、また、前腕筋には、腕橈骨筋、回外筋、円回内筋、方形回内筋等が挙げられる。
【０２２６】
このうち上腕の筋肉についてみると、２関節筋である上腕二頭筋は、肘関節に関する屈曲作用及び前腕に関する回外作用を有しており、特に屈曲については屈筋群の中でも強力な働きをもっていることが知られている。また、上腕三頭筋は長頭が肩甲骨に起始部をもつ２関節筋であり、肘関節における前腕の伸展作用は主に内側頭に属する筋が行い、強力な伸展力を有することが知られている。
【０２２７】
前記した多層構造の効果器を筋肉に対して装着するにあたっては、電極群をシート面の全面に亘って一様に配置するよりは、特定の領域にのみ配置することで効率化を図ることが好ましい。これは、筋肉の場所に応じて電気刺激の影響が異なるからである。尚、「特定の領域」とは、筋肉の筋始点や筋腹、あるいはこれらの近傍領域又は筋始点と終点とに亘る範囲に対応するシート状基材上の領域をいう。例えば、上腕二頭筋を例にすると、この場合には、筋腹に対して配置される電極を陰電極とし、筋始点に対して配置される電極を陽電極とし、対をなす電極を介して低周波電気刺激を与えることによって、皮膚表面から筋肉の伸縮が目視で解るような表層筋に対して刺激を与えることができる。
【０２２８】
また、関節の動作によって巻き込まれてしまう皮膚表面の領域には、電極部を構成する電極群を配置しないことが好ましい。その理由はシート状基材の変形に伴う電極位置の移動が大きいこと、また、上記抵抗パターンの数、圧力や温度の検出素子の数が多い場合には、取得したデータの処理に負担がかかるためである。よって、このような不都合を避けるためには、下記（ｉ）乃至（ｉｖ）に示すように、関節の動作時（屈曲や伸展時等）において圧力が大幅に変化する領域を予め調べておくことが好ましい。
【０２２９】
（ｉ）効果器を対象者に装着する
（ｉｉ）対象者に関節の屈曲や伸展の一定動作を実行してもらう
（ｉｉｉ）（ｉｉ）の動作中における抵抗パターンの抵抗値変化あるいは表面の圧力の情報を取得するとともに、抵抗値や圧力値に関して所定の閾値を設定してこれを越えた位置や範囲を示す場所情報をデータ化してＲＯＭ（リード・オンリー・メモリ）や補助記憶装置（あるいは外部記憶装置）等の記憶手段に格納しておく
（ｉｖ）（ｉｉｉ）で記憶しておいた場所情報を参照して使用電極の配置や、圧力検出を行わない（あるいは検出値を無視すべき）領域を決定する。
【０２３０】
尚、（ｉｉ）においては、対象者にヘッドマウントディスプレイ（ＨＭＤ）等の視覚表示装置を装着して、当該装置上に模倣すべき動作を映し出す方法を用いることが好ましい。つまり、カメラ撮影によって得られる画像情報を動画像として映し出した映像と、これから模倣すべき動作を示す仮想映像を視覚表示装置に表示させ、対象者には、実際の自分の動作を示す映像が仮想映像に重なり合うように動きを追従させることで一定動作を倣ってもらうと、口頭での説明や指示者の動作をまねるよりも能率的である。
【０２３１】
また、効果器を人体に着用して使用できる形状にした場合には、シート状基材のうち電極部を構成する電極群が配置されていない場所に、通気性及び伸縮性を有する材料が用いることが好ましい。つまり、対象者の動作に伴う発熱や発汗に対処する（つまり、蒸れないようにする）とともに動き易さを保証するためには、通気性に富み、伸縮性に優れた材料の使用が望ましい。例えば、ダーリントン伸縮性織地と、水蒸気を通しかつ熱可塑性のフィルムを積層した布地として、リーバンド社の「ダーレックス」（ＤＡＲＬＥＸＸ：商標）や、吸湿性・給水性に優れた綿素材にクーリング性の高いエバール繊維（芯部分にポリエステル、鞘部分にエチエン・ビニルアルコールを用いた複合繊維：株式会社クラレの商標）を交編した機能素材として、ミズノ社の「アイスタッチ」（商標）等が挙げられるが、これらに限らず、スポーツ衣類等の分野で開発されている素材を使用することができる。
【０２３２】
そして、関節及びこれによって駆動される部位の全体に亘って効果器で覆うのではなく、当該関節についての対偶に対応した部分にだけ電極部を有する効果器を部分的に装着して（つまり、対偶毎に電極部を作成するとともに、各対偶の形状的特徴にあった形状の効果器を着用する。）、電気刺激を付与することにより、軽くて着用し易くなる。例えば、前腕の肘関節や手首の関節を例にすると、これらの関節の近傍領域にそれぞれの効果器を装着すれば済むことになる。尚、対偶の形状認識については後述する。
【０２３３】
しかして、本発明に係る制御方法の基本手順を箇条書きにしてまとめると下記のようになる。
【０２３４】
（１）対象者に係る人体の筋肉の配置を示すデータベースを予め構築する
（２）効果器を対象者に装着して、電極群が皮膚表面に接触されるようにセットする
（３）（２）の電極群を構成する電極のうち、陽極側電極として設定された電極への電気刺激信号に係る周波数成分と、陰極側電極として設定された電極から検出される信号の周波数成分との間の差分データを算出することにより、電気刺激により筋から発生した信号の周波数成分を解析し、当該差分データを、ＡＴＰによる多筋線維の動作集合状態信号に係る周波数成分のデータとして保存しておく
（４）皮膚表面の形状及び形状変化を示す情報を取得して形状認識を行う
（５）（１）のデータベースから得られる対象者の筋肉配置の情報と、（４）で得た形状認識結果に基づいて効果器の電極群のうち電気刺激信号を供給すべき電極を選択する
（６）（５）で選択した電極から検出される筋電位情報と、（３）で保存しておいたデータとを比較することにより、筋の動作前にその動作を予測した上で当該電極に対して低周波電気刺激又は干渉波電気刺激のための電気信号を供給する。
【０２３５】
以上で概要説明を終え、下記に示す基本的事項について詳説する。
【０２３６】
（Ｉ）圧力検出を利用した形状認識法
（ＩＩ）骨断面や筋及び脂肪断面の情報を利用したデータベースの構築
（ＩＩＩ）低周波電気刺激法と干渉波電気刺激法
（ＩＶ）ファントムセンセーション
（Ｖ）筋電図を利用した動作予測法
（ＶＩ）低温火傷の防止と刺激出力の調整
先ず、前述したように、図１の形状認識部１ｃに関しては圧力検出を利用した認識方法について説明したが、本法では表面圧Ｐと形状Ｓとの間の数値的関係を予め取得しておく必要があり、そのための初期学習の過程を図３５乃至図４４に従って説明する。
【０２３７】
図３５はシート状基材に行及び列電極シートと感圧層を積層形成して作成した圧力検出用シート１２を円筒状（あるいは円錐台状）に丸めて端部同士をシリコーンやゴム材料、布等を用いて接着した状態を示している（図の斜線部が接着部分１３を示す。）。また、これに限らず図３６に示すように、円筒（圧力検出用シート１２）の側面全体を覆うように接着領域１４を形成しても良い（図に斜線を付して示す部分を参照。）。尚、円筒側面の母線に沿うように縦糸（例えば、麻等の丈夫な工業糸）を接着部あるいはシート基材の内部又は外表面に設けることが好ましい（円筒軸に平行な方向におけるシート基材の形状変化を良好なものにするため。）が、圧力検出用シートの伸張時において電極シートや感圧シートが変形したときに抵抗値や誘電率等が変化する場合があるので、変形に伴う物理量の変化率を予め測定しておき、当該変化率を考慮してデータ補正を行うことが好ましい。
【０２３８】
図３７は圧力検出用シート上における各圧力検出素子を選択するための構成を概略的に示したものであり、マトリックス選択回路１５は、上記した行電極群８ａ、８ａ、・・・に接続される接続線群「Ｘ」と、列電極群１０ａ、１０ａ、・・・に接続される接続線群「Ｙ」を有しており、また、ＣＰＵ１６やメモリ１７（ＲＡＭ又はＲＯＭ）に関するアドレスバス１８及びデータバス１９に接続されている。即ち、ＣＰＵ１６からマトリックス選択回路１５に対するアドレス指定により、圧力検出素子群についての番地指定が為されると、当該番地に対応する行電極群、列電極群のうちの対応する電極がそれぞれ選択され、圧力検出素子の位置が特定される。そして、選ばれた圧力検出素子の検出情報が返されて、データバス１９を介してＣＰＵ１６やメモリ１７において利用されることになる。
【０２３９】
図３８は初期学習時に使用する基準円柱（円筒でも良いが、これを圧力検出用シートに挿入したときに当該シートから受ける力によって形状が容易に変化しないことが必要である。）を示しており、この例では、半径を異にする３つの円柱α、β、γが用意されており、円柱αの半径「ｒ＿α」が三者中で最も小径とされ、その次の円柱βの半径「ｒ＿β」が円柱αの半径よりも大きく、円柱γの半径「ｒ＿γ」が最も大径とされている。
【０２４０】
これらの基準円柱をそれぞれ圧力検出用シート１２（図では円錐台形状をしている。）に挿入した上で圧力分布と形状（基準円柱の半径）との間の数値的関係をキャリブレーションにより取得する。つまり、基準円柱の径が大きいもの程これを筒状の圧力検出用シートに挿入したときの圧迫が大きくなる（表面圧が大きくなる）ことが明らかであり、そのときの圧力検出情報（正確には各圧力検出素子により得られるデータ群）と形状との対応関係を数値情報として得ることができる。
【０２４１】
尚、この例では説明を簡単にするために基準円柱を３個としたので、上記自由曲線を決める上での点の数が少ないが、半径の異なる数多くの基準円柱を用意するか又は１個の基準円柱においてその半径を連続的に可変するための機構を設ける（例えば、空圧や電磁機構等を使った人工筋肉を基準円柱に付設する。）ことによって、補間に必要なデータ数を確保できることは勿論である。また、基準円柱の形状については、円柱の他に４角柱あるいは多角柱等を用いても良い。
【０２４２】
図３９は初期学習の手順例を示すフローチャート図であり、先ず、ステップＳ１で基準円柱（あるいは基準円筒、角柱等）のうちの１つを選択してこれを筒状の圧力検出用シート１２に挿入する。
【０２４３】
次ステップＳ２では、圧力検出用シートにおける各圧力検出素子を上記マトリックス選択回路１５により走査してこれらによる圧力検出情報を取得してメモリ（ＲＡＭ）に格納する。
【０２４４】
尚、この場合、圧力検出素子の数が膨大な数に亘ることに起因してデータ量が増えてしまうという不都合に対処するためには、上記した「人体に関する力学的構造の数値モデル」を活用して必要最小限の数の圧力検出点を選定すると、データ量を大幅に低減することができる。つまり、データ量を削減するために圧力検出素子の数を予め必要最小限にまで少なくしておくことが考えられるが、これでは対象者の個体差（体格差や四肢の長さの違い等）によって圧力検出点を適正な位置に選択できないことになってしまうからである。
【０２４５】
次ステップＳ３では、用意した基準円柱等について、これらを筒状の圧力検出用シートに挿入して圧力分布データを取得する作業が終了したか否かを判断し、終了時には次ステップＳ４に進むが、そうでなければステップＳ１に戻る。
【０２４６】
ステップＳ４では圧力検出ポイント毎に形状Ｓと圧力Ｐとの間の関数関係を得るために、自由曲線や２次曲線等を用いて近似処理を行う。例えば、自由曲線には、ラグランジェの補間曲線、ファーグソン曲線、スプライン補間曲線、べヂエ多項式曲線、Ｂ（ベヂエ）−スプライン曲線等が挙げられ、補間のポイント数に応じて使い分けることが好ましい。尚、補間処理については既知の方法を踏襲すれば良いのでその説明を省略する。
【０２４７】
次ステップＳ５では、前ステップで得られた関数関係を１タイムでＲＯＭにデータとして登録することで初期学習が完了するが、ＥＰＲＯＭ（消去可能プログラム可能なＲＯＭ）やＥＥＰＲＯＭ（電気的消去書き込み可能なＲＯＭ）等を使用することによってその後のデータ補正に容易に対処することができるし、また、前記したようにデータをファイル化して補助記憶（あるいは外部記憶）装置に保存しても良いことは勿論である。
【０２４８】
図４０は圧力検出用シートを対象物に巻き付けた状態で得られる圧力分布情報から形状を認識する手順を示すフローチャート図であり、先ず、ステップＳ１では対象物を筒状の圧力検出用シートに挿入する。例えば、対象者の腕に圧力検出用シートを巻き付ける。
【０２４９】
次ステップＳ２では、圧力検出ポイントを選定後に各ポイントに対応する圧力検出素子によって圧力検出情報を取得した後、次ステップＳ３では、得られた情報をＲＯＭデータと比較する。
【０２５０】
そして、次ステップＳ４では、取得した圧力検出情報に対応する形状データをＲＯＭデータから求めるが、正確に符合するデータがＲＯＭ内にあるとは限らないので、圧力検出情報に近いＲＯＭデータを検索してこれに基づいて予想形状データを算出する。尚、圧力情報から形状を特定するためには多種多様な形状に係るデータを数多く所有してこれをデータベース化しておくことが認識の可否や精度を決定する上で重要である。
【０２５１】
圧力検出用シート（圧力検出手段）を用いて人体の表面形状を認識するにあたっては、関節を如何にして判別するかが問題となるが、圧力検出用シートによって検出される表面圧の変化が大きい範囲を形状認識手段により関節部として認識し、形状モデルとしてのワイヤーフレームモデルの構成データを出力する構成を採用することが好ましい。
【０２５２】
図４１乃至図４４は、上肢を例にしてその形状認識について説明するためのものである。
【０２５３】
図４１は形状認識の手順例を示したフローチャート図であり、スタート時点において前記した初期学習が終了していることが前提となる。
【０２５４】
先ず、ステップＳ１では圧力検出用シートを上肢表面の全体に亘って巻き付けた状態で各検出ポイントにおける圧力情報を取得する。
【０２５５】
次ステップＳ２では、対象者に一定の動作を実行してもらい、その動作が安定して行われるようにする。このためには図４２に示すように、対象者にヘッドマウントディスプレイ（ＨＭＤ）等の視覚表示装置を装着して、当該装置上に模倣すべき上肢の動きを映し出す。つまり、図の画面Ｗ内に示すように、対象者は実際の手を示す映像ＲＨが、模倣すべき動作の仮想映像ＶＨに重なり合うように自分の手を仮想の手に追従させることで動作を模倣する。
【０２５６】
ステップＳ３では、各圧力検出ポイントのうち、予め規定した基準値以上の圧力が検出される領域を選択した後、次ステップＳ４で関節部の判別処理を行う。つまり、図４３に示すように、上肢では手首、肘、肩の部分で対偶動作に伴う圧力の変化が大きいため、この部分に対応する検出ポイントでの圧力情報から関節部を認識することができる。尚、図中に「＋」の丸印で示す位置が関節部を概念的に示しており、また、矢印「Ｒａ」で示す範囲が手首部で圧力変化の顕著な範囲、矢印「Ｒｂ」で示す範囲が肘部で圧力変化の顕著な範囲、矢印「Ｒｃ」で示す範囲が肩部で圧力変化の顕著な範囲をそれぞれ示している。
【０２５７】
次ステップＳ５では、各関節部の認識結果から関節間の長さを求めることでボーンワイヤーモデル（骨格構造をワイヤーフレームモデルとして表現したもの）を作成して形状データを取得する。尚、上肢形状についての認識結果については、例えば、ポリゴンデータとして出力することができる。また、説明は省略するが、脚等の他の部位についても同様の方法を踏襲できることは勿論である。
【０２５８】
関節角度の認識にあたっては、上記のように関節部を認識する際に、対偶の動作に対して圧力変化が新たに加わるポイント又は範囲を探して、当該ポイント又は範囲における圧力検出情報から関節角度を求めることが好ましい。つまり、上記したように関節部周囲の表面は対偶の動作に伴う圧力変化が顕著な場所であり、関節の屈曲角度が大きくなるにつれて次第に圧力が他の場所に比べて大きくなるからである。
【０２５９】
尚、対偶位置の認識にあたっては、図４４に示すいくつかの圧力変化のパターンを満たす検出ポイントを探し出して、その圧力変化から位置を求めることが好ましい。図４４（ａ）乃至（ｃ）に示すグラフ図は横軸に時間ｔ（図４１のステップＳ２で説明した一定動作の開始時点を時間軸の起点とする。）をとり、縦軸に検出された圧力Ｐをとってその時間的変化を概略的に示したものである。
【０２６０】
図４４（ａ）は、グラフ線ｇで示すように動作開始からほぼ一定の傾斜をもって圧力Ｐが上昇していく様子を示しており、また、図４４（ｂ）では、グラフ線ｇ１に示すように、ある検出ポイントでの圧力が最初の立ち上がって飽和した後、別の検出ポイントでの圧力がグラフ線ｇ２に示すように立ち上がって飽和する様子を示している。
【０２６１】
図４４（ｃ）では、複数の検出ポイントでの各圧力を示すグラフ線ｇｉ（ｉ＝１、２、・・・）に示されるように、ある遅延時間をもって次々に圧力が立ち上がった後飽和していく様子が分かる。
【０２６２】
このような圧力の時間的変化に特徴的なパターンが認められる場所を捉えることによって対偶の位置を認識することができ、例えば、手や指、肘、肩等の認識にとって有用である。
【０２６３】
また、捻り動作によって形状が変化する関節部を形状認識手段によって認識する際には、圧力検出手段を構成するシート状基材において捻れに伴って生じる圧力変化から関節角度を求めることが好ましい。例えば、首や腰の回転や前腕の回内・回外動作等において筋肉の捻れを捉えることで形状変化の認識が可能になる。
【０２６４】
尚、皮膚表面の捻れ等に伴って圧力検出シートが体表面から離れることに起因して形成される皺の認識については、急激な圧力変化が２点以上の検出ポイントに亘って連続的に生じた場所を皺とみなすことで処理することができる。但し、形状データを、例えばポリゴンデータとして作成する場合には、多数の皺データを含むことに起因して形状が複雑化し、データ量の増加を余儀なくされることになるので、皺データを除去した表面形状データだけを得たい場合には、ポリゴンデータに関する結合処理を用いる。これはポリゴンメッシュを簡単化して点やラインを間引くことでデータ圧縮を行う時に使用される定番処理であり、エッジのもつ重要度や形状的特徴にどれだけの寄与を持つ要素であるかを考慮した方法や、エッジ除去後の体積変化あるいはエッジ除去後に設定した新頂点ともとの三角パッチの頂点との距離和等を評価量として用いる方法等、各種の方法が知られている。
【０２６５】
以上の形状認識法は、人体の皮膚表面の形状データを取得するのに有効であり、その際、画像認識のための撮像手段が不要であるので、例えば、図２３乃至図２５で説明した対象者の外形形状データを取得する際にも有用である。即ち、上記電極部や圧力検出部をシート状基材の上に多層構造をもって形成した効果器を対象者に着用してもらい、その際に圧力検出部によって得られる皮膚表面の形状データを取得して対象者の外形形状を認識すれば、前記したように対象者の体型形を特定してその筋肉配置を示すデータベースを作成することができる。
【０２６６】
次に、上記事項（ＩＩ）骨断面や筋及び脂肪断面の情報を利用したデータベースの構築について説明する。
【０２６７】
これまでは対象の外形形状や表面状態に関するデータの取得方法に関心をおいて説明してきたが、ヘリカルＣＴ（Ｃｏｍｐｕｔｅｒ Ｔｏｍｏｇｒａｐｈｙ）や、ヘリカルＭＲＩ（Ｍａｇｎｅｔｉｃ Ｒｅｓｏｎａｎｃｅ Ｉｍａｇｉｎｇ）、光ＣＴ（Ｏｐｔｉｃａｌ Ｃｏｍｐｕｔｅｄ Ｔｏｍｏｇｒａｐｈｙ：光断層画像計測＝レーザーを用いた不可視情報の可視化技術の一つであり、例えば、光ヘテロダイン検出法で検出した前方多重散乱直進成分を線形吸収則（Ｌａｍｂｅｒｔ−Ｂｅｅｒ則）に従うものとして投影定理に基づくＣＴアルゴリズムにより画像再構成を行う方法が知られている。）等の断層面撮像（あるいは画像化）手段による画像情報を利用できる場合には、これらを積極的に活用することが好ましい。
【０２６８】
即ち、対象者に基準姿勢（例えば、基本的立位姿勢等）をとってもらうか又は対象部位を撮像台上にギブスで固定してから、ヘリカルＣＴにより得た骨断面と、ヘリカルＭＲＩ等で得た筋や脂肪断面の画像情報を取得するとともに、断面位置の異なる多数の画像データを積層してデータベース化することで対象者に係る静的な人体構造のデータベースを構築する。そして、関節を動作させながら時間軸に沿って対偶を撮影した画像データを取得するとともに、当該関節の角度毎に対偶の形状に係る静的なデータベースを作成することで時間経過を含む動的な人体構造のデータベースを構築する。このデータベースを利用して対象者の筋肉配置を示すデータベースを作成すると、対象者（あるいは対象部位）の外形形状に加えて骨や筋肉、腱等の内部構造や密度分布、電気抵抗値分布を含む、より精密なモデルを得ることができる。
【０２６９】
図４５乃至図４７はそのような手順例を示すフローチャート図である。
【０２７０】
先ず、図４５のステップＳ１で撮影対象となる対偶を選択した後、次ステップＳ２では対偶の状態を考慮して対象部位を撮像台上にギブスで固定する。尚、ここにいう「状態」とは、対偶の動作に影響する筋について関節角度の変化を考慮した状態を意味する。例えば、前腕についていうと、手指計測に関する動作や、尺屈、橈屈、掌屈等の動作を加味して回内・回外運動における断層撮影を行えるように対象部位についての状態設定を行う。その際、筋を緊張させた状態と緊張させない状態とを区別して撮影できるように配慮する。
【０２７１】
次ステップＳ３では、対象部位の長手方向における撮影回数を設定した後、ステップＳ４に進んでヘリカルＣＴによる骨の撮影及びヘリカルＭＲＩや光ＣＴによる骨及び筋の撮影を行う。即ち、骨の輪郭についてはヘリカルＣＴを使って識別し、また、筋肉の輪郭についてはヘリカルＭＲＩや光ＣＴで得た撮像データから、筋と筋との間にある潤滑膜によって識別することができる。尚、後者の場合（断層断面での筋の識別）にあたっては、潤滑膜と筋との色差を利用して画像処理で自動的に識別する方法と、画像処理用ソフトウェアを使って手動又は半手動（例えば、輪郭抽出に係る部分的なオートトレース機能等）で識別する方法があるが、いずれにしても各筋肉の区別が可能である。尚、脂肪の識別については、水プロトン（陽子）と脂肪プロトンでは共鳴周波数にずれがあることを利用して、周波数の違いや位相ずれ等を検出する既知の方法が確立されている。
【０２７２】
次ステップＳ５では、ヘリカルＣＴで得た画像情報において明度差（白い部分と黒い部分との明るさの違い）が顕著な領域を検出する。つまり、骨の輪郭部分が白く、その周囲は黒いので白黒のコントラストが大きいところを輪郭抽出することによって骨の領域を識別することができる。
【０２７３】
ステップＳ６では、ヘリカルＭＲＩ等で得た画像情報において白色領域を抽出した後、次ステップＳ７ではステップＳ５、Ｓ６で得たデータの差分演算によって骨領域を消去したデータを得る。
【０２７４】
そして、図４６のステップＳ８ではヘリカルＭＲＩ等で得た画像情報における白色領域（前ステップで骨領域は除去済みである。）から筋のデータ（輪郭形状の情報を含む。）を抽出して、これに筋の名称を割り振る。その際には、抽出した領域毎に手動で名称を付す方法と、前記人体構造のモデルを利用して自動又は半自動（候補名称の表示支援や、選択肢の呈示等）で筋の名称付けをする方法とがある。
【０２７５】
次ステップＳ９では、各骨や筋の輪郭データから得られる輪郭線の長さをそれぞれ計測して取得した後、ステップＳ１０に進み、ここで対象部位の長手方向における全ての撮影作業が終了したか否かを判断する。そして、終了時にはステップＳ１１に進むが、未終了時には残り撮影回数を１回減らした後で図４５のステップＳ４に戻る。
【０２７６】
ステップＳ１１では、前ステップまでの過程で得られている骨や筋のうちから対象を選択した後、次ステップＳ１２では各断層断面での画像情報についてステップＳ９で得ている各輪郭線の比率を算出する。つまり、ある断層面における骨の輪郭線長を基準して、別の断層面における当該骨の輪郭線長の長さについて比率を割り出す。
【０２７７】
例えば、図４８に示すように、各断層面「Ｓｎ」（ｎ＝０、１、・・・）における輪郭線の形状が三日月形状をしており、それらの輪郭線の長さを「ｌｇｎ」（ｎ＝０、１、・・・）とするとき、図ではｌｇ０＝３０、ｌｇ１＝４０、ｌｇ２＝５０、ｌｇ３＝４０、ｌｇ４＝３０（長さの単位は省略する。）である。よって、ｌｇ０＝３０を基準にすると、これを「１００」とした各長さの比率を得ることができる。
【０２７８】
次ステップＳ１２では、前ステップで得た比率データ（正確にはデータ群）に応じて各断面での輪郭線に対してポリゴンポイント（この時点では輪郭線データは平面的なデータであるが、後述するようにこれを３次元化してポリゴンデータを生成する際にはこのポイントがポリゴンの主要な頂点として利用される。その意味でこれを「メインポリゴンポイント」と呼ぶことにする。）の数（これを「Ｘ０」本と記す。）を指定する。例えば、輪郭線長が長いほどポイント数が多くなるようにする。
【０２７９】
図４８に示す例では、各輪郭線長さ「ｌｇｎ」（ｎ＝０、１、・・・）に対するポイント数を「Ｘ０＿ｎ」（ｎ＝０、１、・・・）とするとき、Ｘ０＿０＝３０、Ｘ０＿１＝４０、Ｘ０＿２＝５０、Ｘ０＿３＝４０、Ｘ０＿４＝３０に規定される（説明の便宜上、ｌｇｎに対するＸ０＿ｎの比率を「１：１」とした。）。
【０２８０】
ステップＳ１３では各断面において尖った部位（角張った部分）を識別し、そのポイント（その数を「Ｘ１」個とする。）を抽出してメインポリゴンポイントとして登録する。
【０２８１】
例えば、図４９に示すような三日月形状では、両端部に鋭角的な部分があり、それらの頂点「ＭＭＰ0」、「ＭＭＰ1」が上記メインポリゴンポイントとして登録されるとともに、この場合のＸ１は２個である。
【０２８２】
そして、図４６のステップＳ１４では輪郭線で囲まれた部分（面）の重心を算出する。例えば、図５０に示すようにスプライン曲線Ｓｐで囲まれた図形に対して、基準点「Ｏ」を設定したときに、当該点と点Ｐ0乃至Ｐn-1上によって構成される多数の三角形群（Ｔｒ0乃至Ｔｒn-1を参照。）によっ対象図形を近似した場合には、各三角形の重心（Ｇ0乃至Ｇn-1）の位置ベクトルを合成することによって面重心（点Ｇ）を求めることができる。
【０２８３】
図４７のステップＳ１５では、前ステップで求めた重心Ｇを中心にして輪郭線を等角で分割する。つまり、分割数を「Ｎ」とすると、「Ｎ＝Ｘ０−Ｘ１」として、重心Ｇから「３６０°／Ｎ」本の直線（放射状の半直線）を引いて、これらと輪郭線（あるいはそのスプライン曲線）との交点を求めて当該交点をメインポリゴンポイントとして追加登録する。
【０２８４】
図５１には点Ｇを中心としてスプライン曲線Ｓｐ（輪郭線）をＮ等分（等角分割）したときの分割線Ｌn_0乃至Ｌn_n-1及びこれらと輪郭線との交点Ｑ0乃至Ｑn-1を示している。
【０２８５】
尚、図４９の場合には、三日月形状（例えば、断層面Ｓ０での形状）について算出される面重心Ｇと、当該重心Ｇを通る「Ｎ＝Ｘ０＿０−Ｘ１＝３０−２＝２８」本の分割線（中心線）Ｌi（ｉ＝２、３、・・・、２９）によって図形分割が行われ（同図（Ｂ）参照。）、各分割線Ｌiと図形の輪郭線（Ｓｐ）との交点「ＭＭＰｉ」（ｉ＝２、３、・・・）が算出されこれらがメインポリゴンポイントとして登録される。
【０２８６】
このように分割数ＮにおいてＸ０からＸ１を差し引く理由は、Ｘ１個の尖った部位に設定されるポイントは予めメインポリゴンポイントとして登録されるので、その数をＸ０（割り当て数）から削減することによって、分割線数を一定化するためである（こうすることによって、各断層面についての割り当て数の比率「Ｘ０＿ｎ／Ｘ０＿０」（ｎ＝１、２、・・・）がくずれないように設定することができる。）。
【０２８７】
上記したとおり、各断層断面での骨又は筋の輪郭線から３次元形状データを作成するにあたっては、当該輪郭線の内部領域の面重心を算出した後、当該面重心から等しい角度間隔をもって放射状に延びる直線群と輪郭線との交点を求め、これらの交点を３次元形状データの基準点として用いると、図４６のステップＳ１３で仮設定されたポイントに対して、さらに詳細な情報内容をもったポイントを付加することができるので輪郭線に対する近似の程度を高めることができる。つまり、ただ闇雲にポイント数を増やすことで高次の近似を行うよりは、ポイントの位置を的確に選定することでデータ処理の負担を軽減することが好ましい（∵面重心Ｇを中心といして当該重心と各ポリゴンポイントとを結ぶ線分の角度間隔が不均等であるよりは、上記点Ｑ1乃至Ｑnに示すように均等な角度間隔をもつようにポリゴンポイントを設定した方がデータ処理が容易であるから。）。
【０２８８】
尚、分割線の数「Ｎ」は面積の大きさに比例して多くすることが高精度化の観点からは好ましいが、上記した説明では、Ｎの決定において面積の代わりにポリゴン頂点数（Ｘ０−Ｘ１）で代用することで処理時間の短縮化を図っている。
【０２８９】
図４７のステップＳ１８ではメインポリゴンポイント間の距離に大きな格差が生じているか否かを判断し（ポイント間の距離の比率あるいはその平均値や分散が、基準値以上であるか否かの比較判断によって行う。）、格差が大きい場合にはステップＳ１９に進み、格差が許容範囲内であればステップＳ２０に進む。
【０２９０】
ステップＳ１９では、メインポリゴンポイント間の距離格差を補正するために、当該格差が問題となるメインポリゴンポイントとこれに隣接するメインポリゴンポイントとの間に補助的なポリゴンポイント（以下、「サブポリゴンポイント」という。）を追加設定する。尚、あるメインポリゴンポイントと他のメインポリゴンポイントとが重なった場合あるいは両ポイント間の距離が隣り合うサブポリゴンポイントの間隔より小さい場合には、先登録のメインポリゴンポイントを優先させる（Ｎ＝Ｘ０−Ｘ１ではなくＮ＝Ｘ０として、重なり合うポイント又は間隔の狭いポイントをスキップして分割線を配置する。）。
【０２９１】
そして、次ステップＳ２０では、ステップＳ２乃至Ｓ１９に至る一連の処理が全て終了したか否かを判断し、そうであれば次ステップＳ２１に進み、未終了時には図４５のステップＳ２に戻る。
【０２９２】
ステップＳ２１では、上記メインポリゴンポイント及びサブポリゴンポイントに基づいて３次元形状データ（ポリゴンデータ）を生成するとともに、これらによって表現される骨や筋等の名称で当該形状データを特定することでデータベースを生成する。
【０２９３】
尚、筋肉の動作をポリゴンデータで表現する際には（形状の動的な表現）、各断層面での面重心（重心群）を通るスプライン曲線を中心線として規定し、当該中心線の形状に合せて筋のポリゴンデータを時間的に変化させる（上記ポリゴンポイントの位置的変化による）と、筋の形状予測が可能になる。
【０２９４】
また、皮膚の変化については対偶の面中心を通るスプライン曲線を規定することで表現することができる。但し、皮膚についてはヘリカルＣＴやＭＲＩ等の撮影時に対象が無負荷状態である場合も想定されるので、このためには圧力検出時の表面形状を考慮して補完処理を行うことが好ましい。そのためには、例えば、下記に示す手順を踏む。
【０２９５】
（１）対象者の骨格及び筋肉のモデルに対して皮膚モデルを付加したモデルを作成する
（２）表面圧力の検出によって皮膚表面の形状データを取得し、これと（１）の皮膚モデルから得られる表面形状とを比較する
（３）皮膚モデルに基づくポリゴンデータと、圧力検出に基づく表面形状のポリゴンデータとが許容誤差の範囲で一致しない場合には、ポリゴンの中心からみてどちらのデータが当該中心から遠いかを判断する（例えば、対偶の中心を基準点「Ｏ」とし、当該基準点を通る半直線が、皮膚モデルに基づくポリゴン図形と交わるポイント（ポリゴンポイント）を「ＰＰ１」、また当該半直線が、圧力検出に基づく表面形状のポリゴン図形と交わるポイントを「ＰＰ２」とするとき、線分「ＯＰＰ１」の長さと、線分「ＯＰＰ２」の長さを比較する等。）
（４）皮膚モデルに基づくポリゴンデータと、圧力検出に基づく表面形状のポリゴンデータとが許容誤差の範囲で一致しない場合には、両者の面的な広がりに関して相対的にどれだけ圧縮又は伸張されているかを比率計算する（つまり、線分「ＯＰＰ１」の長さと、線分「ＯＰＰ２」の長さとの比率を、任意のポイントについて計算することで圧縮率（又は伸張率）を求める。）
（５）（４）で求めた比率に基づいて、両モデルのデータについての大きさがほぼ同じになるように伸縮処理を行うとともに、ヘリカルＣＴやＭＲＩ等から作成した画像についてその基準中心（対偶中心等）に関して圧縮又は伸張処理を行う（画像サイズの変更であり、データ圧縮や伸長ではない）。
【０２９６】
骨や筋、皮膚の名称付けにあたっては、前記した人体構造のモデルから得られる骨や筋等の配置に関するデータからその中心線を求めて、これが図４７のステップＳ２１で得たポリゴンデータにより特定される領域（つまり、骨や筋の推定領域）に含まれるか否かの包含関係をチェックすることで当該ポリゴンデータに係る名称を特定することが好ましく、これによって自動又は半自動での識別処理が可能になる。
【０２９７】
尚、図４５のステップＳ４での撮影において対偶内を部分的に撮影することによって撮影回数及び処理すべきデータ量の削減を図ることができる。
【０２９８】
また、図４６のステップＳ８で画像情報における筋分布の色情報から赤筋と白筋の量的な比率を求めると（断層面では各筋の面積比に相当し、赤色と白色の区別については画像処理（色フィルタをかける）によって色の違いから容易に識別できる。）、赤筋と白筋の比から対象者の反応速度や持久力を推測することができる（赤筋の割合が多い場合には持久力に優れ、白筋の割合が多い場合には瞬発力に優れている。）。
【０２９９】
次に（ＩＩＩ）の低周波電気刺激法、干渉波電気刺激法についての説明に移る。
【０３００】
先ず、低周波電気刺激法について説明する。
【０３０１】
低周波電気刺激法における基本的要素（刺激条件）は下記に示す通りである（諸量については図５２を参照）。
【０３０２】
ｉ）電流の強さ（強度Ｉ）
ｉｉ）刺激パルスの持続時間（パルス幅Ｔｗ）
ｉｉｉ）刺激パルスの極性と波形
ｉｖ）刺激周波数
ｖ）刺激時間と休止時間（Ｔｓ）
ｖｉ）電極配置
ｖｉｉ）電気刺激による運動単位（ｍｏｔｏｒ ｕｎｉｔ）の発射・動員。
【０３０３】
上記ｉ）に関しては、神経や筋を興奮させるにあたってその細胞膜に電位変化を惹き起こすために一定以上の電流強度、電気的エネルギーが必要となる。尚、強度が強い方が刺激効果は高いが、不快感や痛みも強くなってしまうので、刺激強度の設定にあたっては効果に対するデメリットにも配慮する必要があり、５０ｍＡ（ミリアンペア）以下、例えば、２０ｍＡ〜３０ｍＡ程度が目安となる（医療用具では出力電流２０ｍＡが上限とされている。）。また、上記ｉｉ）の持続時間やｉｖ）の周波数によっても刺激強度（あるいは筋収縮力）が変化することにも注意を要する。
【０３０４】
尚、刺激電流の（時間）平均値がゼロでないと電極と生体との間に分極現象が生じてしまい電気化学的組成の変化をもたらすので好ましくない。そのためには、容量（コンデンサ）や絶縁トランス等を用いて直流成分をカットする。
【０３０５】
ｉｉ）の持続時間に関しては、ある一定の電気的エネルギー量の供給により神経や筋を興奮させるための刺激パルスの持続時間をどのように規定するかが問題となる。つまり、刺激波形のエネルギー量は刺激強度Ｉとパルス幅Ｔｗとの積に基づいて決定されることになるが、一般にはパルス幅が長くなると筋収縮の度合が強くなるが、逆に痛みが誘発されることにもなるので、約０．２〜０．３ミリ秒程度のパルス幅が用いられる。
【０３０６】
尚、刺激強度とパルス幅との間には一定の関係があることが知られており（所謂、強さ・時間曲線）、刺激強度が弱い場合には長いパルス幅を必要とし、逆に刺激強度が強く場合には短いパルス幅で筋収縮が行われる。そして、パルス幅をどんなに長くしても一定の電流値以下では収縮が起らなくなったときの電流の強さを「基電流」といい、基電流の２倍の電流値でのパルス幅を時値（クロナキシー）と呼んでいる。
【０３０７】
また、神経や筋に加える電気刺激については持続時間の短いパルス電圧（あるいは電流）が用いられ、例えば、筋の収縮量を制御するための刺激パルス列についての変調法には、振幅変調（ＡＭ）、パルス幅変調（ＰＷＭ）、周波数変調（ＦＭ）が用いられる。
【０３０８】
上記ｉｉｉ）の極性については、刺激電極を陰極にし、不関電極（電気刺激を加えた際に刺激を感じない電極）を陽極に設定する。これは、一般に筋収縮を誘発させる刺激電流が陰極下の場合に筋肉の興奮性が高いためであり、負性パルスが使用されることが多い（例えば、負性矩形波では、刺激による痛みや不快感を極力低減するとともに、筋の疲労が少なく滑らかで力強い筋収縮力が得られる。）。
【０３０９】
また、神経、筋を興奮させるには電気的エネルギー量とともに、一定以上の急峻な電流変化が必要であり、波形の立ち上り時の傾き（あるいは傾き角）がある程度大きいことが必要となる。即ち、大きな電流量であっても非常にゆっくりと電流強度を上げていった場合には順応によって刺激とはならないためである。尚、この「傾き」（刺激パルスの立ち上がりの角度であり、図５２の「θ」を参照。）は刺激波形に関係しており（理想矩形波では９０°である。）、波形には負性矩形波、補正単相・指数関数的増減波等が挙げられる。
【０３１０】
上記ｉｖ）の周波数に関して、「低周波」をどの範囲の周波数とするかについての考え方がさまざまであるため明確な定義はないが、２００Ｈｚ以下あるいは１００Ｈｚ以下の範囲が目安となる。尚、単発の電気刺激では筋が単収縮を起こすだけで強縮は起さない。つまり、強縮を惹き起こすには一定以上の刺激周波数が必要であり、正常では１５Ｈｚ程度以上である（１５Ｈｚ以下の低周波で刺激すると筋の震えが出現する。）。また、刺激周波数が筋の反応帯周波数に近づく程収縮力も強くなる。
【０３１１】
上記ｖ）の刺激時間と休止時間については両者の比が問題となり、デューティーサイクル（あるいはデューティー比）の決定にあたっては筋の疲労を考慮する必要がある。つまり、刺激時間が長すぎると容易に筋肉疲労を来してしまうので、刺激強度や周波数の如何にもよるが、一般的には刺激時間に対して１：１以上の休止時間が必要である。尚、筋の持久力を増加させたい場合には休止時間を基準値よりも短くし、最大筋力を増加させたい場合には十分な休止時間を確保しながら強い筋収縮を引き出す刺激条件を設定する等の配慮が必要になる。
【０３１２】
上記ｖｉ）の電極配置については、筋収縮を誘発するために電気刺激を与える場合、最も反応性の良い部位（所謂、運動点＝ｍｏｔｏｒ ｐｏｉｎｔ）を考慮して決定することが好ましい。つまり、上記した形状認識部１ｃによって対象部位の形状や動きを認識した後、筋配置を特定して筋への刺激領域が運動点を含むように決定する。尚、運動点は神経が筋に入り込んでいく部分であり、その場所は神経筋接合部付近と考えられている。
【０３１３】
また、神経本幹が皮下に浅在している場合にはその部位も良好に反応するので、この運動点への刺激に付与するための電極配置や形状については通電方法を単極法、双極法、グループ刺激法に分けて規定することが好ましい。
【０３１４】
つまり、単極法では、大きな不関電極を、目的筋から離れた、筋の少ない部位に置き、小さな刺激電極（関電極）を運動点上に置いて刺激を行う。例えば、上記した効果器では、多数の電極によって占有面積の大きい不関電極を構成し、少数の電極によって占有面積の小さな不関電極を構成すれば良い。
【０３１５】
双極法では、大きさが同じで比較的大きな電極によって運動点を挟むようにして配置するか、あるいは陰極を運動点に置いて刺激が行われる。上記した効果器では、使用電極の数及び位置の選択によって刺激面積の設定を容易に行うことができる。
【０３１６】
グループ刺激法では、ある筋群全体を収縮させるような電極配置が使用される。
【０３１７】
低周波電気刺激の場合に２極１組の電極対を対象部位に複数設置する際には、各組の電極間の導通予想ライン（刺激電流路の予想ライン）が互いに交わらないように電極配置を設定することが望ましい。これは、導通予想ラインが交わる場所の筋には強力な干渉波刺激が加わってしまう虞があり、このような事態の回避が必要となる。尚、電極間の導通予想ラインの算出や判定にあたっては、電極群のうち隣り合う電極の抵抗成分を検出するとともに、前述した断層面情報から得られる対象部位の内部構造のデータについて骨、筋肉、脂肪の材質に応じた抵抗率（あるいは比抵抗）を設定して計算することができる。例えば、上記したように電極２、２、・・・が格子配置となっている場合には、ある電極を中心として選んだときにその周囲に８つの隣り合う電極が存在するが、これらの電極の１つと中心の電極とを選択して一定の微弱電圧をかけることで両者間の抵抗値を知ることができ、このような検出を別の電極についても同様に行うことによって表面の抵抗値分布を知ることができる。また、内部構造に係る抵抗分布については、各構成部位（骨、筋、腱、脂肪等）に係る既知の抵抗率を前記した断層面解析の結果に基づいて各部に設定した上で、有限要素法等、既知の方法を用いて求めることができる（導通予想ラインは、対をなす電極を特定したときに、両電極の間で最も抵抗の小さい場所を通る経路として算出される。）。
【０３１８】
上記ｖｉｉ）について、通常の随意筋収縮では負荷の増加に伴って発射する運動単位が、より細い神経線維に支配されている遅筋（ｔｙｐｅＩ）筋線維から、より太い神経線維とその支配している速筋（ｔｙｐｅＩＩ）筋線維の発射として参加してくる（これを「Ｓｉｚｅ ｐｒｉｎｃｉｐｌｅ」と呼んでいる。）。
【０３１９】
しかし、電気刺激ではこれとは逆に刺激強度を上げていくと、より太い神経線維から刺激されるため、筋疲労が早期に出現してしまうことになるので、これに対する配慮（つまり、適切な刺激条件の設定）が必要となる。
【０３２０】
また、電気刺激の場合、太い神経線維（つまり、運動神経線維）から刺激され、より細い感覚神経線維がその後から刺激されることになるため、運動神経に係る刺激の閾値が通常の範囲であれば、表在感覚神経刺激による痛みを与えることなく、運動神経を刺激して筋収縮を得ることが可能である。
【０３２１】
次に、干渉波電気刺激法について説明する。
【０３２２】
この方法は、皮膚抵抗を無視できる程度の中周波電流（例えば、４０００Ｈｚと３９００Ｈｚ）を使用して周波数差に相当する低周波数電流（ビート電流）を発生させるものであり、皮膚抵抗による電力損失が少なく、しかも人体内部組織の部分で低周波を発生させる方法であるため対象部位について十分な電気刺激のエネルギーを供給であるという利点がある（低周波電流を皮膚に対して直接的に流す場合には電気的エネルギーの大部分が皮膚の電気抵抗によって消費されてしまうため、内部組織に十分なエネルギーを供給するのが困難である。）。また、皮膚表面から筋肉の収縮が目視で認識できないような内層筋（例えば、指伸筋等）に対しては干渉波電気刺激が望ましい。
【０３２３】
干渉波電気刺激の基本的要素（刺激条件）は、下記の通りである。
【０３２４】
ｉ）刺激周波数差（干渉周波数）
ｉｉ）ビート電流のエネルギー量。
【０３２５】
筋刺激に適した刺激条件としては、例えば、パルス波形として立体導体波（三極干渉波）を用いた場合に、出力電流１００ミリアンペア以下、搬送周波数５０００Ｈｚ、干渉周波数０．５〜２００Ｈｚ程度である。また、皮膚刺激に適した刺激条件としては、例えば、パルス波形を負性矩形波とした場合に、出力電流５０ミリアンペア以下、搬送周波数４０００〜４１５０Ｈｚ、干渉周波数１５０Ｈｚ以下である。
【０３２６】
効果器における使用電極の設定にあたっては、干渉波刺激の様子を実験的に調べる方法もあるが、干渉波シミュレーター（携帯電話のアンテナ設計等において使用する電磁界シミュレーターを用いて、人体の骨や筋等に対して異なる伝搬係数を決定して複数の運動点に関する干渉波のシミュレーションを行うための装置）を利用することが好ましい。
【０３２７】
その際、骨や筋、脂肪、腱に対してそれらの材質毎に減衰率（σ）を設定して、皮膚表面のうち各部までのエネルギー損失が少ない場所に電極を設定すると、エネルギー消費に関して効率的である。つまり、前述した方法により対象部位の断層面情報から内部組織の情報（つまり骨や筋等）が得られている場合には、それらの３次元形状データ（ポリゴンデータ）に対して材質毎の減衰率を設定を設定して干渉波シミュレーションを行うことができる。
【０３２８】
図５３は皮膚表面Ｓ上の任意の点Ｓａ（Ｓａｘ、Ｓａｙ）と、骨や筋等の部位ＸＰの内点Ｐとが距離「ｒ」だけ離れており、部位ＸＰに対して波の減衰率「σ」を与えたときのエネルギー損失「Ｌ」が、点Ｓａの位置座標の関数Ｌ（Ｓａｘ、Ｓａｙ）として表されることを示している（図では単に「Ｌ（Ｓａ）」と記す。）。つまり、内点Ｐについて関数値の小さい場所をシミュレーションで探知することで当該場所に位置する電極あるいは当該場所に最も近い場所の電極を使用電極として設定することができる。尚、使用電極が決まると、そのときの対象部位の内点Ｐが干渉ポイントとなるが、このポイントついては関節の駆動角度毎に設定することが望ましく、また、指筋群のように密集した筋配置については対偶の長手方向においてそれぞれ異なる位置に干渉ポイントを設定することが望ましい（干渉ポイントが狭い領域に集中しないようにするため。）。
【０３２９】
さらに、各部位ＸＰに対してそれらの材質毎に物質密度あるいは質量密度（ρ）を設定して、皮膚表面のうち各部までの波の位相ずれが小さい場所に電極を設定すると、干渉波刺激にとって効果的な場所を選定することができる。尚、物質密度（ρ）は材質に応じた質量密度（単位体積当たりの質量）から決定できる。
【０３３０】
そして、部位ＸＰの内点Ｐでの理想的な干渉波を取得するためには、電極位置の設定についてさらに下記の手順（１）乃至（４）を採ることが好ましい。
【０３３１】
（１）骨や筋、脂肪、腱に対してそれらの材質毎に波の反射率（α）及び屈折率（β）をそれぞれ設定する
（２）骨や筋、脂肪、腱に対してそれぞれ設定された減衰率（σ）や物質密度（ρ）に基づいて上記したように最小エネルギー損失の場所、最小の位相ずれの場所での電極候補から複数の電極を選び出す（図５４には表面Ｓ上に電極候補位置を３点「Ｓａ１」、「Ｓａ２」、「Ｓａ３」で示す。）
（３）（２）で選んだ電極位置から対象部位ＸＰの内点Ｐに向かう搬送波の時間的推移について当該部位ＸＰによって反射あるいは屈折される波形を平面的にシミュレートすることで、当該平面上における干渉波の周波数分布を求める
（４）内点Ｐでの理想的な干渉波が得られるまで、（２）及び（３）の手順を繰り返した後、最終的に使用電極位置を決定する。
【０３３２】
尚、上記（２）の手順において、始めから減衰率（σ）や物質密度（ρ）を考慮した電極候補の選出を行うよりも、初期の試行では減衰率（σ）や物質密度（ρ）を考慮せずに電極候補を選出した後、電極場所についての見通しがある程度ついてから減衰率（σ）や物質密度（ρ）の設定及びこれに基づく電極候補の選定を行う方がシミュレーションの効率アップにとって好ましい。
【０３３３】
また、こうして設定された電極についてはその出力電流値を変更して対象部位ＸＰの内点Ｐへの電気刺激の出力設定を行うことができるように装置を構成することが好ましい（干渉波における電極配置について適切な出力設定を行うためである。）。また、筋の配置領域は時として関節の動作（例えば、前腕の回内・回外動作等）によって著しく変化することがあるので、各関節の動作毎にシミュレーションを行う（つまり、上記動的な人体構造のデータベースに基づいて時間経過を考慮した干渉波シミュレーション）を行って使用電極位置を決定することが好ましい（∵対象部位の時間経過を含まない静的な干渉波シミュレーションだけでは、関節動作に対応した筋への干渉波のための電極を適切に設定することが難しいから。）。尚、その際、対偶の運動状態を示す情報（ポリゴンデータ等を含む。）と、これに対する理想的な刺激領域及び使用電極位置の情報とを関連付けたデータベースを構築しておいて、データの再利用に供する（例えば、各使用電極に優先度を付けたり、使用電極に不具合が生じた場合に別の候補電極で代用するための属性データをデータベースに含める等。）ことが望ましい。そのために、複数の電極候補を取得した後、これらの位置情報をメモリや補助記憶装置等の記憶手段に保存しておき、実際に電極として使用する際に電極候補の位置情報を読み出してこれらの中から電気刺激の対象領域に適した電極を選択する。
【０３３４】
上記の説明ではシミュレーションの対象を内点Ｐとしたが、内層筋群への電気刺激にあたっては、点ではなく一定（幅）の周波数帯域をもつ領域内の面を対象としてシミュレーションを行う方法と、さらには面を拡張して立体（例えば、球や円柱、多面体、角柱等）を対象にシミュレーションを行う方法とが挙げられ、次元数が大きくなる程に処理が複雑となる。
【０３３５】
前者の方法では、図５３や図５４において点Ｐの代わりに対象部位ＸＰの内部や境界に面を設定すると、複数の筋肉を同時に刺激したり、電気刺激に反応する領域の面積を大きくした干渉波シミュレーションを実現することができる。
【０３３６】
面設定にあっては、下記の場合分けが挙げられる。
【０３３７】
ａ）隣り合う筋への同時刺激
ｂ）動作させたくない筋と動作させたい筋とが入り交じっている場合。
【０３３８】
先ず、ａ）では、複数の筋のうち隣り合う筋が同時に刺激されるように対象面を設定すると、ある特定の筋とその付近の筋への干渉波シミュレーション結果を同時に得ることができる。例えば、隣接する複数の筋肉の同時動作については、隣接する筋の各中心に対して理想的的な周波数帯域の領域が複数の筋の範囲に含まれるように広範囲な干渉領域に対して電気刺激を行うことが好ましい。そして、実際の電気刺激を付与する際には、干渉波シミュレーション時に設定した面を刺激領域とし、当該刺激領域に適した電極候補の中から使用電極を選定して電気刺激を付与する。
【０３３９】
また、ｂ）では対象部位ＸＰについて内点Ｐを設定するか又は面を設定するかを領域毎に使い分けることで動作させたくない筋を刺激対象から除外できる。つまり、動作の対象となる筋と不動作の対象となる筋とが入り交じっている場合には、これらの筋に対して点状又は面状の刺激領域を複数設定して干渉波シミュレーションを行うことが好ましい。
【０３４０】
内層筋群への電気刺激にあたっての電極選定時に、一定（幅）の周波数帯域を有する刺激領域を立体領域として干渉波シミュレーションを行う場合、つまり、シミュレーションを対象部位ＸＰ内の立体領域について行う場合には、例えば、複数の面を「ＳＰｉ」（ｉ＝１、２、・・・）とするとき、「ΣＳＰｉ」（Σはｉについての和を表す。）として決まる境界面とその内部領域における任意の点によって立体領域が決定されて干渉波シミュレーション結果が得られるので、電気刺激に反応する領域（の体積）を面の場合よりもさらに大きくとることが可能になる。
【０３４１】
尚、低周波電気刺激と干渉波電気刺激について同時にシミュレーションを行う場合には、低周波刺激の予想領域を想定した上で干渉領域についてのシミュレーションの演算処理を行うことで、対象部位について同時に低周波及び干渉波による刺激を与える場合の刺激領域について適切な電極位置の選定が可能になる（つまり、低周波電気刺激については前記したように導通予想ラインの交錯を避けることが必要である。）。
【０３４２】
以上のような干渉波シミュレーションに必要な計算量は、候補電極の数が多くなるほど顕著になってくるのでシミュレーターの負担を軽減して処理速度を向上させるためには、下記の手順を踏む。
【０３４３】
（１）駆動対象となる関節の角度及び当該関節角度毎の３次元形状データ（骨、筋、脂肪、腱等のポリゴンデータ）と、当該データに基づく各部位の位置関係に基づいて得られる使用電極位置の候補を関連付けてこれを予めデータベースとして構築しておく
（２）ある関節角度について（１）で候補とした使用電極位置で実際に電気刺激を行う
（３）（２）の電気刺激によって得られた対偶の駆動量と、（１）の関節角度に対応する対偶の駆動量とを比較して両者の一致、不一致を判別する
（４）関節角度を変えて（２）及び（３）の手順を繰り返す。
【０３４４】
尚、手順（３）において駆動量の一致が認められない場合には、その相違を補正するために必要な情報（例えば、関節角度の差）をデータ登録したり、あるいは使用電極位置や刺激強度を変更する等して誤差補償を行う必要がある。つまり、対偶の駆動が期待した通りにいかない場合には、該電極への刺激出力の設定値を変更する方法又は別の電極を選定する方法、両者を併用する方法が挙げられる。
【０３４５】
また、シミュレーションに際してポリゴンデータの簡略化も有効であり、例えば、筋についてはポリゴンの断層断面での重心を結合したワイヤーフレームモデルを採用したり（例えば、筋形状を示す３次元ポリゴンデータをワイヤーフレームデータとして簡略化した上でデータベースを作成する。）、各種パラメータを各筋について一律に規定する等の方法が挙げられる。
【０３４６】
複数の筋を同時に動作させるにあたっては、これらの筋に対して電気刺激を同時に与える方法の他、収縮させたい筋に対して、筋が収縮して再度伸張しない時間内で時分割で電気刺激を付与する方法（干渉波の刺激領域として点状又は面状又は立体状の領域を設定して、これらの領域に対して時分割処理で電気刺激を付与する方法）がある。後者の方法では、皮膚の同じ場所について継続的に電気刺激を与えない点で低温火傷の防止にも有利である。
【０３４７】
尚、筋への電気刺激を実際に与えるときに、干渉波領域を点状、面状又は立体状として干渉波が発生されることになるが、筋が密集した部位において各筋を独立に動作させる場合には、お互いの干渉波領域が重複により干渉し合わないように設定する（∵領域が重複した場合には筋の独立動作が保証されない虞があるから。）。また、重力による対偶や関節への影響を無視することはできないので、関節や対偶の位置と、これらにかかる重力との関係を考慮した上で筋肉への刺激を行うことが好ましい（例えば、関節や対偶がある位置や姿勢をとった場合に、これらにどの位の重量がどのような角度をもって加わっているかに応じて、伸筋と屈筋との間の収縮比率を変化させることで、重力の影響によって対偶の動作が不安定になる等の不都合を防止することができる。）。
【０３４８】
次に（ＩＶ）ファントムセンセーションの説明に移る。
【０３４９】
この場合には、表層の皮膚に対して複数の電極から周波数の異なる電気刺激が与えられ、これによって触覚を付与することができる。つまり、上記した干渉波電気刺激を皮膚表面について応用することで対象者への触覚刺激の呈示が可能となる。
【０３５０】
その際、下記の呈示が挙げられる。
【０３５１】
ｉ）点触覚
ｉｉ）線触覚
ｉｉｉ）面触覚
ｉｖ）点触覚から面触覚への変化あるいはその逆
ｖ）点触覚や線触覚の移動による刺激領域の変化。
【０３５２】
先ず、ｉ）点触覚の呈示については、上記した干渉波領域を点（干渉ポイント）として表現することで実現することができる。つまり、対象部位の内点Ｐを指定して干渉波電気刺激を付与することで皮膚表面を尖ったもので押されたかのような感触が得られる。
【０３５３】
また、ｉｉ）線触覚については、干渉波領域を線（つまり、複数の干渉ポイントを繋いだ線）として表現することで実現することができる。この場合には皮膚表面を板の側面で押されたかのような感触が得られることになる。
【０３５４】
ｉｉｉ）面触覚については、干渉波領域を面として表現することで実現できるので、この場合には皮膚表面を平面板や曲面板で押されたかのような圧力の感触が得られることになる。
【０３５５】
尚、ｉｉ）やｉｉｉ）については、線が点の集合であり、面が線の集合であることに着目すると、点状（あるいは線状）の干渉波領域を高速に移動させる（対象者の点や線の移動を意識させない程短い時間内での領域移動が必要である。）ことで線触覚（あるいは面触覚）を擬似的に実現する方法が挙げられる。
【０３５６】
ｉｖ）については、干渉波領域を点から面、あるいは逆に面から点へと時間的に変化させることで電気刺激を付与する。例えば、仮想現実を利用したゲーム機等への適用において仮想空間上に視覚的に現出される敵からの攻撃を受けたときに着弾の感触を実現するためには、刺激領域を点から円に拡げた後、円から点へと刺激領域を収束させるという様に電極位置及び範囲を選定して電気刺激を対象者に与えれば良い。
【０３５７】
ｖ）については、点触覚や線触覚をゆっくりと移動させたときの刺激領域の変化に伴う感触である。例えば、仮想現実を利用したゲーム機等への適用において仮想空間上に視覚的に現出される敵からの攻撃を受けて刀傷を負ったときの感触を実現するために、点状あるいは線状の刺激領域を、刀傷に沿う線（刀先の軌跡を示す。）に沿って移動させることで電極位置又は範囲を時間的に変更しながら電気刺激を対象者に与えれば良い。
【０３５８】
尚、干渉波とファントムセンセーションについて同時にシミュレーションを行う場合には、ファントムセンセーションにより刺激を付与する部位に係る干渉波の刺激領域と、干渉波による内層筋への刺激領域を想定した上で干渉領域についてのシミュレーションの演算処理を行うことで、対象部位への同時刺激を付与する場合の刺激領域について適切な電極位置の選定が可能になる。
【０３５９】
次に（Ｖ）筋電図を利用した動作予測法について説明する。
【０３６０】
従来よりＥＭＧフィードバックを用いた電気治療器が知られているが、筋トーヌス（ｍｕｓｃｌｅ ｔｏｎｕｓ：筋緊張＝力を加えていない筋を被動的に伸展した際に被験者が感じるかすかな抵抗であり、その本質は筋伸展反射であるが、中枢神経、末梢神経（α、γ−系）、筋自身によって複雑な制御を受けている。尚、筋伸展反射とは筋が受け身で引き伸ばされると、その筋が脊髄からの運動指令によって収縮する結果、力を出す反射のことをいう。）についてはＡＴＰ（アデノシン三リン酸）による多筋線維の動作集合信号とともにＥＭＧに現れるため、筋の動作前に動作予測を行うことができなかった。尚、この「多筋線維の動作集合信号」については、中枢神経から末端神経に指令が伝達される際の神経興奮により筋トーヌスによる筋細胞の収縮に伴って発生されるものであり、体調や体格、血糖値、筋疲労の度合によって変動する。
【０３６１】
そこで、以下のようにして筋トーヌスを予測するとともに、この予測筋トーヌスと表面筋電図との間の相対関係を習得することで、動作予測を行う。
【０３６２】
図５５はＡＴＰによる多筋線維の動作集合信号を取得するための装置例を示しており、この場合の対象部位は上腕二頭筋（２関節筋であり、肘関節に関する屈曲作用及び前腕に関する回外作用を有する。）である。
【０３６３】
上腕二頭筋の筋腹には陰（極）電極２０Ｎと検出用電極２０Ｄが配置され、筋始点には陽（極）電極２０Ｐが配置され、これらの電極には電気刺激発生装置（例えば、医療用低周波治療器等）２１からの電気刺激信号が供給される。
【０３６４】
筋電波形の周波数解析のために用意された２チャンネルＦＦＴ（高速フーリエ変換）アラナイザー２２には、陽電極２０Ｐへの供給信号が点Ａで分岐した後でアンプ２３を介して第１ｃｈ（チャンネル）のＦＦＴ部２２ａに送出される。また、検出電極２０Ｄからアンプ２４を経て取り出された信号が第２ｃｈのＦＦＴ部２２ｂに送出されるようになっており、各ＦＦＴ部の出力は後段の差分演算処理部２２ｃに送出され、その出力はディジタル化の後、コンピュータ２５に取り込まれる。尚、コンピュータ２５からＦＦＴアラナイザー２２にはタイミング制御等に必要な制御信号が送出されるようになっている。
【０３６５】
電気刺激発生装置２１から陽電極２０Ｐに送られる信号の周波数成分がＦＦＴ部２２ａで解析されるが、これは電気刺激の原信号となるものであって筋の応答を含んではいない。
【０３６６】
筋への電気刺激によって発生した周波数成分は、検出用電極２０Ｄからアンプ２４を介してＦＦＴ部２２ｂで解析される。つまり、電気刺激に対する筋の追従により生じた応答についての周波数解析が行われる。尚、ＦＦＴ処理にかかる時間（ΔｔF）については対象者の筋トーヌス発生からその後の動作変化に要する時間（ΔｔM）よりも充分に短い（ΔｔF＜ΔｔM）ことが必要である（∵「ΔｔF＞ΔｔM」では、周波数解析処理が筋の動作に追い付かないので、動作前の予測が不可能となってしまう）。
【０３６７】
差分演算処理部２２ｃにおいて周波数成分を帯域毎に比較して差分演算の結果得られる差分データは、電気刺激発生装置２１による電気刺激によって筋から発生した周波数成分（原信号の周波数成分を除いたもの）である。つまり、電気刺激の付与だけによって筋が収縮すると、筋トーヌスは出ずに筋細胞の収縮に伴う電位情報を取得することができるので、筋トーヌス及び筋細胞の収縮を含むＥＭＧのデータから上記差分データ（ＡＴＰによる多筋線維の動作集合信号）を差し引くことで筋トーヌスの予測が可能となり、また、対象者個人の筋特性を取得することができる。
【０３６８】
図５６はＦＦＴ後の周波数特性を概略的に示したグラフ図であり、横軸に周波数（ｆ）、縦軸にパワースペクトルをとって筋電図周波数特性図を示したものである。
【０３６９】
図中のグラフ曲線ｇａは、対象者の意志で筋を動かしたときの筋トーヌスと筋細胞収縮を含む信号のＦＦＴ解析結果を示しており、グラフ曲線ｇｂは、筋に電気刺激を付与したときの上記差分データから得られる筋細胞収縮の周波数成分を示している。つまり、グラフ曲線ｇａに示す周波数成分からグラフ曲線ｇｂの成分を差し引いた部分（図の斜線参照。）が筋トーヌスに係る成分を表している。
【０３７０】
よって、筋に付与する電気刺激の波形を各種変更した場合の上記差分データに基づいて予測される筋トーヌスと、そのときの表面筋電図の電位情報とを関連付けて記憶・学習を行うことで、動作予測が可能となる。
【０３７１】
以上に説明した、予測に必要な筋電位情報と、これを用いた予測手順とを箇条書きにしてまとめると次のようになる。
・必要な情報
Ａ）対象者がその意志により筋肉を動作させるときに得られる筋電位情報（これを「第１の筋電位情報」と呼ぶことにする。）
Ｂ）電気刺激だけによって筋肉を動作させたときに得られる筋電位情報（これを「第２の筋電位情報」と呼ぶことにする。）。
・予測手順
（１）筋への電気刺激信号についての周波数成分の解析結果と、電気刺激によって筋から発生した周波数成分を含む信号の周波数成分の解析結果とを取得して両者を比較することで、電気刺激により筋から発生した周波数成分（つまり、筋トーヌスによらない筋細胞の収縮に伴う波形成分）だけを差分データとして抽出する。つまり、第２の筋電位情報に係る周波数成分には、電気刺激信号の周波数成分が含まれているので、これを除去した成分を求めることで、ＡＴＰによる多筋線維の動作集合信号の成分を抽出できる。これが筋収縮時における運動単位の特性を反映した成分に相当する。
【０３７２】
（２）（１）で得た差分データと、対象者がその意志で筋肉を動かそうとしたときに得られるの筋電波形（筋トーヌスを含む。）のデータから筋トーヌスに係る成分を取得する。つまり、第１の筋電位情報に係る周波数成分には、筋トーヌスに係る成分と、ＡＴＰによる多筋線維の動作集合信号の成分とが含まれているので、後者の成分である差分データを第１の筋電位情報から差し引くことで、筋トーヌスに係る周波数成分のデータだけを取得し、これを筋トーヌスの予測用データとする。尚、この事は、ＦＦＴ解析による周波数特性において、「第１の筋電位情報であるＥＭＧデータ（の周波数成分）−ＡＴＰによる多筋線維の動作集合信号（の周波数成分）＝筋トーヌス（の周波数成分）」という概念的図式に対応する。
【０３７３】
（３）電気刺激信号波形を各種変更して（１）及び（２）の手順を繰り返すことで、その時々の予測筋トーヌスと表面筋電波形データとの関係を記憶・学習する（例えば、ニューラル・ネットワークの利用等。）。つまり、上記第１の筋電位情報と第２の筋電位情報に基づいて算出される筋トーヌスの予測用データを、対象者の動作パターン（対象筋の収縮に伴う対偶動作の経時的変化を示す一連のデータ群であり、形状を示すポリゴンデータや、関節角度等のように形状変化を示すパラメータの他、手の内転、外転や腕の回内・回外等の、形状や動作を特定するためのデータ等が含まれる。）と関連付けて記憶・学習する。
【０３７４】
（４）対象者が自分の意志で筋肉を動かそうとしたときの筋電波形データを得た場合に、（３）で習得した情報に基づいて予測される筋トーヌスから筋の動作を予測する。つまり、対象者がこれから筋肉を動作させようとする時の筋電位情報について周波数解析を行い、その結果から算定される筋トーヌスの成分データを取得する。そして、当該データと、既に記憶・学習済みの予測用データとを比較・照合して、両データの一致が認められたときに、該データに対応した筋肉の動作パターンを予測する。こうして、実際に筋が動作する前に筋トーヌスを知ることで筋の動作予測が可能となる（筋トーヌスのデータは動作毎に異なった特徴をもつ周波数成分を含んでいるため、例えば、対象者がこれから重量物を握ろうとしているといった動作をその直前に知ることができる。）。
【０３７５】
上記のように、電極から検出した第１の筋電位情報に係る周波数成分から、ＡＴＰによる多筋線維の動作集合状態信号の周波数成分を取り除いた情報に基づいて筋トーヌスの予測を行うことができる（これは上記制御手段１ｄが担当する。）。
【０３７６】
尚、上記周波数解析部１ｆに相当するＦＦＴ部２２ａ、２２ｂによって得られた差分データを取得できると、電気刺激を受けたときの筋細胞の収縮の様子が分かるようになるので、対象者個人の筋特性に係るデータを取得することができる。例えば、筋特性には、赤筋と白筋の比率に起因する筋の反応速度や持久力の相違等が挙げられるが、筋細胞の収縮に伴う筋電位情報は、体調や血糖値、筋疲労の度合によって変動することに注意を要する（つまり、この成分が筋肉の状態によって一定でないことが筋の動作予測を難しくしている原因であり、本発明は、上記第１の筋電位情報から当該成分を除去するための処理を行うことで、運動単位の特性について考慮することなく筋の動作予測を行うことに着眼点をおいたものである。）。
【０３７７】
また、図５５では電気刺激発生装置２１の電極を例に挙げて説明したが、前記した効果器においては電気刺激に適した電極位置（つまり、上記干渉波シミュレーターによって選定した位置）と同じ位置の電極から筋電波形データを取得して、陽極側電極及び陰極側電極からのそれぞれ信号について周波数解析を行うことで筋トーヌスについて予測することができる。つまり、効果器の電極２、２、・・・を通して筋に電気刺激を付与するだけでなく、これらの電極を、筋電図情報の取得にも活用することができる（つまり、筋電計として利用できる。）。
【０３７８】
図５７は入出力インターフェイス部（入出力選択処理部１ｅ）における切換制御によって電極を通した電気刺激と、電極からの筋電位情報の取得を時分割処理で行う場合の構成例を概念的に示したものである（図では１個の電極についての等価回路を示す。）。
【０３７９】
電気刺激発生部１ｂから供給される信号は出力バッファ２６を経た後、切換部２７（図にはスイッチの記号で示す。）から電極部１ａｄに送出される。また、電極部１ａｄから切換部２７を経た信号は、入力バッファ２８を介して筋電位情報取得部２９（上記ＦＦＴ部２２ａ、２２ｂや差分演算処理部２２ｃを含む。）に送られる。切換部２７の状態は図示しない制御部から送られてくる制御信号によって規定され、切換部２７において出力バッファ２６側が選択された場合に電極を通して電気刺激が筋に付与され、また、入力バッファ２８側が選択された場合に電極を通して筋電波形の情報が取得される。
【０３８０】
このように時分割処理によって入出力（Ｉ／Ｏ）を切り替えることで同一の電極を通した電気刺激の付与と筋電位情報の取得を行うことができるように構成するか、あるいは、両者について同じ使用電極がかち合わないようする（選択した電極を通した電気刺激の付与と、当該電極を検出電極とした筋電位情報の取得とが同時刻に同じ電極で行われないように別々の電極を選択する）必要がある。
【０３８１】
尚、前記した方法で選択した使用電極を通して電気刺激を筋に与えたときに得られる実際の動作が、干渉波シミュレーションの結果と食い違う場合には使用電極の位置を別の候補電極（候補登録済の電極）の位置に変更することで実際の動作に最も合う電極位置を選択することが好ましい。
【０３８２】
しかして、周波数解析から予想される筋トーヌスから筋の動きを事前に把握して、動的な人体構造のデータベースを用いることにより、対象者の関節について次の動作を予測できるようになる。つまり、上記した手順（４）における予測筋トーヌスと表面筋電波形データとを関連付けるデータベースと、前記した動的な人体構造のデータベースとを用いることで対象部位について次に予測される動き（モーション）に関する情報を、表面筋電波形データに基づいて習得することが可能になる。
【０３８３】
尚、壁を手で押すときのように、対偶が動作することなく筋収縮が起る場合（静止性収縮）の筋の仕事量や、対偶の動作速度の取得に際しても筋電図の利用が好ましく、例えば、対象者にヘッドマウントディスプレイ等の視覚表示装置を装着してもらった上で当該装置上に映し出される動作を模倣したり、あるいは関節を装具で一時的に固定しあるいは一定の負荷をかけながら筋電パターンについてのデータを取得する。
【０３８４】
次に、（ＶＩ）低温火傷の防止と刺激出力の調整について説明する。
【０３８５】
絶縁接触抵抗による低温火傷を防止するには、下記の方法が挙げられる。
【０３８６】
ａ）ある一定の時間以上に亘って同一の電極を使用して電気刺激の付与が行われないように使用時間を制限して、使用電極を変更する方法
ｂ）皮膚表面の絶縁抵抗が予め決められた閾値以上である場合には、その電極からの刺激強度を低下させるか又は当該電極の使用を禁ずる方法。
【０３８７】
先ず、ａ）における使用電極を変更については、使用する電極間の抵抗値が閾値以上になった場合に、当該電極とは別位置の電極を代替電極として用いる方法が挙げられる。例えば、電極の配置パターンの変更による方法、時分割処理によって各電極への通電時間の持続を制限する方法（例えば、隣接する電極群同士が同時に使用されることがないように、各グループに属する電極群を時分割処理に従って使用すると、皮膚表面の同じ領域に対する長時間の刺激付与を避けることができる。）、両者を併用する方法等がある。
【０３８８】
また、ｂ）は、使用する電極群の間の抵抗値を測定して、抵抗値が閾値以上になったことが判別された場合には、当該電極を通して付与される電気刺激の強度を低下させるか又は当該電極の使用を強制的に停止させる方法である。尚、強度低下については対象者の体脂肪率や痛覚に対する個性を把握した上で刺激に対する許容範囲を考慮する。また、電極の使用禁止については、一時的な禁止（解除条件つきの禁止。例えば、電源の再投入時に絶縁抵抗値を測定してこれが閾値以下であれば使用禁止を解く等。）と、永続的な禁止（解除条件をつけることのない禁止）とが含まれる。そして、使用禁止を宣告された電極については、これを除外して再度干渉波シミュレーションを行って候補電極を選定して禁止電極の代替電極を見つけるか、あるいは既に登録済の候補電極のうちから優先度等を考慮して禁止電極の代替電極を選定する必要があり、これらによって安全な使用電極を保証することができる。また、低温火傷は、あかや汚れ等の付着によって皮膚抵抗が大きくなったときにも起こり易いので、皮膚の絶縁抵抗が許容範囲を越えたことを検出したときに、電気刺激を禁止することが必要である。
【０３８９】
人体の脂肪はそのまま電気刺激における絶縁抵抗となるため、刺激による不快感に深く関わっている（つまり、低温火傷は抵抗値に比例したジュール熱に起因するため。）。
【０３９０】
そこで、対象者をその体脂肪率によってクラス分けを行い、各クラスごとに刺激強度と不快感についての有限段階（例えば、快、不快を含む５段階）での主観評価を行うことが好ましい（∵体脂肪率だけでは脂肪の付き方に関する情報が不足するため。）。この結果を利用することで低温火傷に関する個体差を識別して対象者に許容される最大刺激値を越えないように電気刺激の出力調整を行うことができる。
【０３９１】
図５８及び図５９は人体の脂肪と刺激による主観評価法の一例を示すフローチャート図であり、本例では評価対象部位を屈曲筋である上腕二頭筋とし、刺激強度に対する評価を被験者の両腕の間隔で表現する（例えば、両腕をまっすぐ前にのばした状態において、不快感が大きい程、両腕で手先間隔を大きくしてもらう。）ことにより、刺激がどの程度の不快感として被験者に認識されているかを個々の被験者について相対的に認識する方法を用いている（∵不快感についての絶対的指標による評価は困難であるため。）。
【０３９２】
先ず、図５８のステップＳ１では、被験者に前記した効果器を上腕に装着してもらった上で、当該効果器を通して複数の電気刺激（例えば、３段階の刺激Ａ、Ｂ、Ｃ）を個々に与えた後、次ステップでＳ２では、被験者に、刺激Ａを最小、刺激Ｂを中間、刺激Ｃを最大の刺激としてそれぞれ認識してもらうように口頭で説明を行う。
【０３９３】
次ステップＳ３では被験者に刺激Ｃがどの程度の不快感を与えるかを両腕の間隔によって表現してもらった後、ステップＳ４に進んで刺激Ａ、Ｂ、Ｃをランダム（無作為）に選択して被験者に対して付与する。尚、その回数は所定回数（例えば、１５回程度）とする。
【０３９４】
次ステップＳ５では、各刺激（Ａ、Ｂ、Ｃ）がどの程度の不快感を与えるかを被験者に両腕の間隔で表現してもらう。これによって、刺激Ｃの場合における両腕の間隔を基準とした相対値により評価する（例えば、刺激Ｃの場合の間隔を１００とし、刺激Ｂが５０％、刺激Ａが２５％等というように百分率で評価する等）。
【０３９５】
次ステップＳ６において所定回数に亘るランダムの刺激付与が終了したか否かを問い、終了時には図５９のステップＳ７に進むが、未終了時にはステップＳ４に戻る。
【０３９６】
ステップＳ７では刺激Ｃの強度をこれまでより１段階増加して被験者に付与し、次ステップＳ８で当該刺激がどの程度の不快感を与えるかを被験者に両腕の間隔で表現してもらう。
【０３９７】
そして、ステップＳ９では被験者の両腕が所定の間隔以上に大きく開いたか否かを判断し、そうであれば次ステップＳ１０に進み、そうでなければステップＳ７に戻る。
【０３９８】
ステップＳ１０では、効果器の電極群から得られる筋電位情報に基づいて変化の顕著な場所を特定してその上位から所定数の電極位置及び筋電パターンのデータを取得する。
【０３９９】
こうして被験者について刺激強度と不快感との関係を個人的特性として相対的に評価することができる。
【０４００】
尚、被験者の筋肉に対する最大印加電流（最大強度）の決定にあたっては、被験者に筋肉を緊張させてもらって刺激電流を徐々に増加していき、下記の状態が認められた時の電流値に安全率を乗じた値又はこれが刺激電流の上限値（最大電流値あるいは許容上限値）を越える場合には当該上限値とする。
【０４０１】
ｉ）関節が電気刺激によって降服したとき
ｉｉ）痛いと感じたとき。
【０４０２】
尚、関節角度と電気刺激との関係については、一定の刺激出力を対象筋に与えてそのときの関節角度に応じて出力設定や調整を行えば良い。
【０４０３】
また、人体構造モデルのデータベースに基づいて算定される対象者の人体内部の抵抗分布と対象者の体脂肪率から予測される抵抗値とが整合しない場合に絶縁抵抗値の修正を行うことがモデル誤差の低減にとって望ましい。例えば、対象者の体脂肪率に比べて算定した抵抗値が小さい場合には、各抵抗値データに乗ずる比例係数を変更する（無修正時の比例係数の値は１である。）。
【０４０９】
【実施例】
図６０乃至図６５は本発明に係る装置について実施の一例を示すものであり、本装置は、Ｉ／Ｏ処理部（機構部）１００、電気刺激発生部３００、統括制御部５００を備えている。
【０４１０】
各構成要素は以下の通りである（括弧内は符号を示す）。
※Ｉ／Ｏ処理部（１００）
・電極部（１０１）
シート状基材に配置・形成された多数の電極群から構成され（図４乃至図８参照。）、後述の電気刺激発生部３００から電極選択部１０４を通して送られてくる電気信号が各電極に供給される。
・圧力及び温度検出部（１０２）及び素子選択処理部（１０３）
シート状基材において上記電極部１０１の上層に配置・形成された多数の検出素子部（図９乃至図１２参照。）から構成され、圧力や温度の検出素子の選択については素子選択処理部１０３におけるマトリックス処理により行われる。
・電極選択部（１０４）
出力系統については、電気刺激発生部３００からの信号群（例えば、「チャンネル数」×２個の信号）が出力用セレクタ１０４ｏに送出され、ここで制御信号「Ｓｅｌ＿ｏ」を受けて選択される電極に対してバッファアンプ１０４ｂを通して出力信号が供給される。尚、バッファアンプ１０４ｂについては、制御信号「Ｓｃ」によって出力電流の設定が行われるように構成されている。また、入力系統については各電極からの信号が入力用セレクタ１０４ｉに送出され、ここで制御信号「Ｓｅｌ＿ｉ」を受けて選択される情報が後述する筋電検出部２０１に送出される。
・３軸位置検出部（１０５）
対象者についての基準原点位置を検出するための検出部であり、対象者に関する座標系（３次元直交座標系、あるいは円筒座標系や関節構造に応じた軸座標系等）の設定に必要とされる。尚、検出方法には光学式、磁気式、ジャイロセンサー方式等が知られている。
※筋電検出部（２０１）
電極選択部１０４において指定された電極で取得した筋電位情報をアナログ信号からディジタル信号にＡ／Ｄ変換して認識する部分であり、前記した周波数解析部１ｆやＦＦＴ部２２ａ、２２ｂの機能を有している。尚、この筋電検出については常時行っても良いし、タイマー割り込み等によって規定のタイミングで行ったり、あるいは検出命令を受けたときに行うようにしても良い。また、Ａ／Ｄ変換後のデータ処理形態については、筋電波形のパルスについて規定される閾値を越えたときの単位パルス当たりの立ち上がり数だけを用いても良いし、また筋電波形を実効値変換のために時間積分した積分筋電図（Ｉｎｔｅｇｒａｔｅｄ ＥｌｅｃｔｒｏＭｙｏｇｒａｍ：ＩＥＭＧ）から得たり、あるいは筋電波形をＦＦＴ処理した後スペクトル互換法を適用してデータを得る等、各種の方法がある。
※形状認識部（２０２）
圧力及び温度検出部１０２から素子選択処理部１０３を経て入手される圧力検出情報を受けて圧力分布情報から表面形状を認識する部分である。尚、圧力検出情報の処理については全てマトリックス処理を経て行われる。
※個性認識部（２０３）
筋電検出部２０１及び形状認識部２０２からの情報に基づいて対象者の個性を認識する部分であり、認識結果は後述の統括制御部５００に送られる。尚、「個性の認識」においては、上記した方法の他、筋電図情報の周波数解析により所定帯域（例えば、２００〜３００Ｈｚ）の成分から識別する方法や、複数の対偶動作について得られる筋電位情報又は形状認識情報を合成して、この構成情報と個人情報との差分情報を個性として認識する方法（音声認識技術を筋電位情報等に応用した方法）等を用いることができるが、その方法の如何は問わない。
※電気刺激発生部（３００）
低周波刺激発生部３０１及び干渉波刺激発生部３０２が必要なチャンネル数に応じて設けられており、これらは後述する統括制御部５００の指令下に置かれている。
※対偶位置認識部（４０１）
３軸位置検出部１０５、筋電検出部２０１、形状認識部２０２からの情報に基づいて対偶の位置情報を認識する部分である。つまり、３軸位置検出部１０５からの情報を原点位置情報とし、対象者に特定の動作を実行してもらいながら、その時々の形状認識結果や対偶運動との関係、運動時の筋電位情報や運動速度等の関係に基づいて対偶の位置を認識する。
※骨格筋配置認識部（４０２）
対偶位置認識部４０１からの情報に基づき、人体構造モデル処理部６０２からの３次元の人体形状データ（対偶内の骨や筋肉等の配置をポリゴンデータでモデル化したもの。）を利用して骨や筋肉の配置を認識する部分であり、その際、人体形状データについてはモニター装置４０３上に表示して視覚的に認識できるようにすることが好ましい。
※統括制御部（５００）
制御の主要な機能を司る部分であり、これには以下の要素が含まれる。
【０４１１】
・刺激選択機能処理部（５０１）
電気刺激の種類を選択するための制御信号を電気刺激発生部３００に送出する部分である。即ち、電気刺激を行う対象部位の筋肉について皮膚からの深さに応じて低周波電気刺激を行うか又は干渉波電気刺激を行うかを規定し、低周波については２つの電極を１組みして電気刺激を発生させ、干渉波については、例えば、４つの電極を組みして電気刺激を発生させる。尚、電気刺激にあたっては、前記したように干渉波シミュレーター（又はその簡易版干渉波シミュレーター）からの解析結果が利用される。
【０４１２】
・Ｆ．Ｓ．（ファントムセンセーション）刺激機能処理部（５０２）
皮膚表面への擬似的な刺激を付与する際に使用する電極の選定及び刺激強度等の信号パターンを生成する部分であり、この場合には前記したように干渉波が使用される。また、後述する力触覚パターン認識部５０５からの指令情報に応じた触覚呈示を行うために、制御出力を後述の使用電極変更処理部５０７に送出する。尚、ファントムセンセーション刺激によって機能的電気刺激が妨害される場合には、使用電極パターンの配置を変更して、速やかに他の電極パターン配置を採用する必要がある。
【０４１３】
・低温火傷防止処理部（５０３）
絶縁接触抵抗による低温火傷を防止するために設けられており、一定時間以上に亘って同じ電極を使って電気刺激が付与されないように使用電極のパターンを適時に変更したり、あるいは皮膚表面の抵抗値が閾値以上である場合や電極の汚損状態がひどい場合に電極の使用を禁止する。そして、そのための制御指令を後述する使用電極変更処理部５０７に送出する。尚、図５８、図５９で説明したように筋電検出部２０１からの情報を参照し（図には破線で示す。）、個人差に応じて刺激に対する不快感の発生防止対策を講じることが好ましい。
【０４１４】
・筋配置変化への追従機能処理部（５０４）
筋の移動に伴って使用電極の配置等を変化させるための制御を行う部分である。例えば、前腕の回内・回外動作時における筋肉の移動はとても大きいため、その際の筋肉移動を考慮して使用電極配置を時間的に変更する必要がある。よって、骨格筋配置認識部４０２によって得られる筋肉配置の情報、筋電検出部２０１から予測された対偶の駆動量（前記予想筋トーヌスに基づく動作予測を参照。）、そして、実際の対偶位置情報の変化に基づいて対偶の駆動量をリアルタイムで認識し、対偶間の位置関係に応じて使用電極位置及び電気刺激波形を変化させる。
【０４１５】
・力触覚パターン認識部（５０５）
対象部位にどのような力触覚を付与するかを認識する部分であり、その制御法は力触覚フィードバック制御（つまり、指令値に対してどれだけの力触覚が付与されているかを認識して両者についての差分処理から求まる誤差がゼロとなるような制御）に依る。その際には、後述の人体構造モデル処理部６０２からの３次元人体形状データを利用する。尚、認識結果については、ファントムセンセーション刺激機能処理部５０２、筋配置変化への追従機能処理部５０４等に送出される。
【０４１６】
・選択処理／出力設定処理部（５０６）
使用電極の選択（電極位置、使用又は不使用若しくは禁止の選択を含む。）、選択した電極を使って電気刺激を付与するのか又は当該電極を使って筋電位情報を取得するのか、あるいは電極をグランドにとるか等を規定する。また、選択電極を通して電気刺激を付与する場合の出力設定を行う。尚、選択処理／出力設定処理部５０６からセレクタ部５１２を介して上記制御信号Ｓｅｌ＿ｏ、Ｓｅｌ＿ｉ、Ｓｃが各所に送出されるが、当該セレクタ部５１２が完全に選択処理／出力設定処理部５０６の支配を受けるのは、後述する使用電極変更処理部５０７からの指令を受けないことが条件とされる（つまり、セレクタ部５１２にとっては当該指令が優先されるということである。）。
【０４１７】
・使用電極変更処理部（５０７）
各種の条件によって使用電極を変更するための部分であり、その出力はセレクタ部５１２に送出される。尚、本処理部の出力はセレクタ部５１２に対して優先的に作用する。また、使用電極の変更指令については、上記したファントムセンセーション機能処理部５０２、低温火傷防止処理部５０３、筋配置変化への追従機能処理部５０４、後述の信号補正処理部５０８等から送られてくるが、各処理部のもつ機能に応じて各種の電極配置パターンが用意されている。
【０４１８】
・信号補正処理部（５０８）
骨格筋配置認識部４０２から得られるデータに基づいて予期される力触覚フィードバック（状態）と、実際に起きている力触覚フィードバック（状態）とを比較し、両者が一致するか又は差異が少なくなるように、刺激パターン設定部５１０や使用電極変更処理部５０７に対して補正信号を送出する部分である。
【０４１９】
・個性適応機能処理部（５０９）
対象者の個性認識用データに基づいて電気刺激波形をその標準形から変化させるための機能をもった部分であり、個性認識部２０３からの情報を受けて刺激パターン設定部５１０に対して制御信号を送出する。
【０４２０】
・刺激パターン設定部（５１０）
統括制御部５００内の各部（５０２〜５０４、５０８、５０９）からの指令に基づいて刺激パターン（前記した刺激条件を異にする各種の刺激信号パターン）に関する制御信号を電気刺激発生部３００に送出し、波形等を変化させる。
【０４２１】
・簡易版干渉波シミュレーター（５１１）
干渉波シミュレーター６０３で得られた情報を用いて電極候補の位置、対偶駆動量等を予めデータベース化しておくことで干渉波シミュレーションの使用頻度を低減するためのものであり、使用時には、予想される対偶駆動量と、実際に候補電極を通して電気刺激を付与したときの対偶駆動量とが一致するか否かを判別して使用電極を決定する。
※入出力処理部
力触覚パターン（指令パターン）を設定したり、３次元ポリゴンデータを画像表示する等に必要な一切の装置を含む。例えば、コンピュータの入力装置６０１（キーボードやマウス等）、モニター装置４０３やヘッドマウントディスプレイ装置、プリンタ等の出力装置の他、前記した人体構造（モデル）のデータベースに係るデータ入出力処理に必要な手段も含まれる。
※人体構造モデル処理部（６０２）
前記したモデル構築／計算処理部１Ａｅに相当する。尚、モデル構築にあたっては、既述したとおりヘリカルＣＴやＭＲＩ等によって得られる断層面情報から対象部位の骨格や筋肉配置のデータを得ることが可能である。
※干渉波シミュレーター（６０３）
前記したように干渉波電気刺激の付与に際して使用電極候補の選出等に用いられる。
【０４２２】
尚、力触覚呈示装置への適用にあたっては、仮想現実空間におけるオブジェクトの生成や力触覚呈示の指令に関する処理が必要となり、例えば、ゲーム機への応用にあたっては図６４、図６５に示すように前記の構成要素に加えて、ゲーム機７００内に以下の要素が必要になる（括弧内の数字は符号を示す。）
・ステレオ映像出力部（７０１）
ストーリー展開に従ってステレオ映像（信号）を出力する部分であり、当該映像信号は、例えば、ヘッドマウントディスプレイ（あるいは３Ｄスコープ）等の視覚表示装置７０２に送出されて対象者に提供される。尚、効果音等の音声信号についても視覚表示装置に組み込まれたヘッドフォンやスピーカー等の音声出力手段を使って対象者に提供されるが図示は省略する。また、ゲームの進行に必要なデータ等は図示しない情報記録媒体によって提供される。
【０４２３】
・力触覚発生の命令部（７０３）
力触覚の発生について命令を下す部分であり、ストーリー展開において力触覚の呈示を必要とする場面で力触覚の呈示命令を発してこれを上記力触覚パターン認識部５０５に送出する。これにより力触覚パターン認識部５０５で呈示命令の内容が解釈される。また、命令部から対偶位置認識部４０１に送出される指令によって力触覚呈示の対象部位について対偶位置の認識及び位置情報の取得が行われる。
【０４２４】
次に機能的電気刺激による力触覚呈示装置の制御方法について、図６６乃至図７１に示すフローチャート図に従って説明する。
【０４２５】
図６６は制御全体の流れについてその大筋を示したものであり、対象者の人体構造に関するデータベースに対して対偶の動作や筋電図によるデータをリンクしたデータベースが構築済であること、そして、該データベースを利用した干渉波シミュレーションの結果がやはりデータベース化されていることが前提となる。尚、関節角度に応じた人体の骨格や筋肉の配置データについては前記したようにヘリカルＣＴにより骨格データを取得し、またヘリカルＭＲＩ等で筋肉配置のデータを取得してこれをデータベース化しておく。
【０４２６】
処理の開始にあたっては、先ず、電極部１０１や圧力及び温度検出部１０２を含む効果器や、ヘッドマウントディスプレイを対象者に装着する。その際、電気刺激の対象部位について皮膚全体が効果器によってしっかりと被覆される必要がある。
【０４２７】
ステップＳ１では基準面圧力情報の取得にあたって必要なキャリブレーション処理（その詳細については図６７で説明する。）を行う。
【０４２８】
そして、次ステップＳ２では関節角度による面圧力変化についての情報取得にあたって必要なキャリブレーション処理を行う（これは求心性収縮及び遠心性収縮に関する認識のために必要となり、その詳細については図６８で説明する。）。
【０４２９】
次ステップＳ３では筋の仕事量についてのキャリブレーション処理（その詳細については図６９、図７０で説明する。）を行った後、次ステップＳ４では前ステップＳ１乃至Ｓ３で得られたデータと、前述した人体構造モデルのアルゴリズムに従って生成されたデータ、そしてヘリカルＣＴやＭＲＩ等の情報を用いて生成したポリゴンデータ（関節の動きによって筋の中心線や形状がどのように変化するかの認識に必要とされ、その詳細については図４５乃至図４７やその関連説明を参照。）についてリンクを行う。尚、この場合にリンクの基準とするのは関節角度であり、その際、対偶の形状から推測される関節角度も考慮する。
【０４３０】
そして、ステップＳ５で力触覚呈示命令を解釈する。つまり、当該命令は力触覚発生の命令部７０３から力触覚パターン認識部５０５に送出されてその内容が解釈される（上位の命令は下層の構成部にいくに従って、より低次元の命令に還元されていく。）。
【０４３１】
例えば、力触覚パターンの設定条件については運動パターン、スピード、仕事量等が挙げられ、これに基づいて皮膚表面の刺激位置、強度、種別（例えば、刀傷や着弾等を受けたり、バウンドしたといった感触の違いに応じた類型）等、各種のパラメーターが規定される。尚、刺激位置の設定にあたってはゲーム機の場合にそのストーリー展開の応じて指定する必要があり、そのためのデータを予め情報記録媒体に記憶させておく。
【０４３２】
次ステップＳ６では対偶位置と重力との関係や荷重を考慮して制御を行う。つまり、力触覚発生の命令部７０３から対偶位置認識部４０１に送出される指令によって対偶位置の認識が行われるが、その際に対象者への重力の影響が問題になる。
【０４３３】
制御方法には以下の３種類が挙げられる（ここでは、一つの関節を複数の筋肉で駆動する場合、例えば上腕等を想定する。）。
【０４３４】
・関節とその駆動を補助する筋肉の収縮に関するデータベースを予め構築しておき、電気刺激に際しては当該データベースを用いて、常に対をなす筋肉を特定してこれらを緊張させるように制御する方法
・関節及び対偶の位置と重力との関係についてのデータベースを予め構築しておき、電気刺激に際しては当該データベースを用いて、重力に対する関節位置に応じて対をなす筋肉を特定してこれらを緊張させるように制御する方法
・荷重（例えば、手にもった重量物等による。）に対しては、刺激対象とする駆動筋をデータベースに追加しておき、電気刺激に際しては当該駆動筋を選択してこれを緊張させるように制御する方法。
【０４３５】
次ステップＳ７では力触覚呈示命令に従って使用電極の選択や刺激条件の設定が行われて電気刺激が付与される（その詳細については図７１で説明する。）。
【０４３６】
図６７は基準面圧力情報の取得（キャリブレーション）に関するフローチャート図であり、先ず、ステップＳ１では、磁気センサー等を用いた３軸位置検出部１０５からの情報に基づいて基本となる対偶位置の情報を取得した後、次ステップＳ２では３軸位置検出部１０５からの情報を用いて視覚表示装置（ヘッドマウントディスプレイ等）上の映像において対象者の目視位置（視点）情報を取得する。
【０４３７】
次ステップＳ３では、前ステップＳ１で得た対偶位置に基づいて当該対偶からその末端部までの人体部位についての３次元ポリゴンデータを、前記人体構造モデルのアルゴリズムに従って生成した後、次ステップＳ４に進んで当該ポリゴンデータに対応する映像を上記視覚表示装置７０２に映し出す。
【０４３８】
そして、ステップＳ５において、対象者にはポリゴンデータの映像に対して自身の対偶（の映像）が重なり合うように動作状態を模倣してもらう。尚、その際には音声によるナビゲートも同時に行う。
【０４３９】
次ステップＳ６では映像の重なり具合から動作の模倣が完了したか否かを判断し、完了時には次ステップＳ７に進んでそのときの圧力分布を効果器の圧力検出素子群の情報から取得した後ステップＳ８に進むが、模倣が完了していない場合にはステップＳ５に戻る。
【０４４０】
ステップＳ８では、前ステップＳ７で取得した数値データを基準値（例えば、百分率表示を採用した場合の「１００」）として記憶・登録し、これに基づいて基準面圧力情報のデータベースを作成する。
【０４４１】
図６８は関節角度に応じた面圧力情報の取得（キャリブレーション）に関するフローチャート図であり、ステップＳ１、Ｓ２については図６７のステップＳ１、Ｓ２と同じである。
【０４４２】
ステップＳ３では、前ステップＳ１で得た対偶位置と関節角度に基づいて当該対偶からその末端部までの人体部位についての３次元ポリゴンデータを、前記人体構造モデルのアルゴリズムに従って生成した後、次ステップＳ４に進んで当該ポリゴンデータに対応する映像を視覚表示装置７０２に映し出す。
【０４４３】
そして、ステップＳ５において、対象者にはポリゴンデータの映像に対して自身の対偶（の映像）が重ね合わさるように動作を模倣してもらい（尚、その際には音声によるナビゲートも同時に行う。）、次ステップＳ６では映像の重なり具合から動作の模倣が完了したか否かを判断し、完了時には次ステップＳ７に進んでそのときの圧力分布を効果器の圧力検出素子群の情報から取得した後ステップＳ８に進むが、模倣が完了していない場合にはステップＳ５に戻る。
【０４４４】
ステップＳ８では、図６７のステップＳ８で作成した基準面圧力情報のデータベースのデータを基準として前ステップＳ７で得た圧力情報のうち変化の大きな場所及び圧力値を取得する。例えば、基準面圧力情報と比較した場合に圧力検出値の変化に富む場所を上位から所定箇所だけ選び出してそれらの位置及び圧力値を記憶・登録し、これに基づいて関節角度と当該角度についての面圧力情報のデータベースを作成する。
【０４４５】
そして、ステップＳ９では関節の全ての角度について上記ステップＳ３乃至Ｓ８の処理を終了したか否かを判断し、未終了時にはステップＳ３に戻って関節角度をこれまでの値とは変更した上でステップＳ３乃至Ｓ８の処理を繰り返す。
【０４４６】
このようにして関節角度をパラメータとして変化させたときの面圧力の変化を基準面圧力に対する相対値として取得することができる。
【０４４７】
図６９及び図７０は筋の仕事量による筋電（パターン）情報の取得（キャリブレーション）に関するフローチャート図であり、図６９のステップＳ１乃至Ｓ６については図６８の各ステップと同じである。
【０４４８】
図７０のステップＳ７では、効果器の電極部１０１を通して筋電位情報を取得できるように設定した後、筋電波形のパターンを複数回に亘って取得する。
【０４４９】
そしてステップＳ８では、前ステップで得た情報が既に記憶・登録済か否かを判断し、そうであればステップＳ９に進み、未登録の場合にはステップＳ１０に進む。
【０４５０】
ステップＳ１０では、前ステップで得たデータについて平均処理を施した上でこれを基準値（例えば、百分率表示を採用した場合の「１００」）として記憶・登録し、これに基づいて無負荷時の筋電パターン情報に係るデータベースを作成する。そして、次ステップＳ１１に進んで対象部位に所定の負荷を与えるための装具を装着した後、ステップＳ３に戻る。
【０４５１】
ステップＳ９では、ステップＳ１０で作成した無負荷時の筋電パターン情報を基準として前ステップＳ７で得た筋電位情報のうち変化の大きなデータを取得する。例えば、無負荷時の筋電パターンと比較した場合に電位の変化が顕著なデータを上位から所定の数だけ選び出してそれらを記憶・登録し、これに基づいて負荷時の筋電パターン情報に係るデータベースを作成する。
【０４５２】
このようにして対象部位の無負荷時における筋電位情報に対して負荷（付与）時の筋電位情報を相対値として取得することができる。
【０４５３】
図７１は電気刺激の発生による力覚呈示の手順例を示すフローチャート図である。
【０４５４】
先ず、ステップＳ１で関節の駆動角度を設定した後、次ステップＳ２で圧力検出に基づいて対偶形状を把握する。そして、ステップＳ３では対象部位の３次元ポリゴンデータ（あるいは断層解析結果に基づく内部構造の形状データ）を生成した後、次ステップＳ４では駆動させたい筋肉について刺激領域（前記した干渉波シミュレーションの説明における点や面、立体）を設定して電気刺激を付与する。
【０４５５】
これによって、ステップＳ５では実際に対偶が駆動されるが、次ステップＳ６では前ステップＳ１で設定した駆動角度が達成したか否かを判断し、達成時にはこれまでの処理を終了するが、当該角度が達成されない場合にはステップＳ２に戻る。
【０４５６】
尚、人体の動作において上腕や手部では多数の筋が互いに協調しているため、全ての筋について干渉波電気刺激を個別に付与するのは非常に困難である。そこで、対偶を動作させている筋の配置に応じて低周波電気刺激と干渉波電気刺激（触覚呈示のためのファントムセンセーションを用いた刺激法を含む。）を併用することが望ましい。
【０４５７】
図７２は筋動作予測のための筋電位情報の取得と記憶・学習処理の一例を示すフローチャート図である。尚、上記した装置構成において特に直接的な関係のある部分は、入力用セレクタ（１０４ｉ）と筋電検出部（２０１）、出力用セレクタ（１０４ｏ）と電気刺激発生部（３００）である。
【０４５８】
先ず、ステップＳ１は準備処理のための分岐処理であり、ステップＳ２乃至Ｓ４のうち、未終了とされるステップに進む。
【０４５９】
ステップＳ２では対象者に効果器（例えば、スーツ形状に形成したもの等）を着用した上で解剖学的立位姿勢をとってもらい、効果器において皮膚表面に接触する面に皺ができないようにセットした後、前記（Ｉ）圧力検出を利用した形状認識法に従って対象者の表面形状データを取得する。尚、このデータ取得処理が必要である理由は、後述のステップＳ３やＳ４だけでは、筋肉の隆起等に伴う形状変化を動的に把握することができないからである。
【０４６０】
また、ステップＳ３では、前記（ＩＩ）骨断面や筋及び脂肪断面の情報を利用したデータベースの構築において説明したようにヘリカルＣＴやＭＲＩ、光ＣＴ等を使った断層断面のデータ取得及びデータベース化の処理を行う。
【０４６１】
ステップＳ４では、対象者の人体構造に関するデータベースを構築する（前記した力学的構造の数値モデルの構築方法に拠る。）。
【０４６２】
ステップＳ５では全ての準備処理が終了したか否かを判断し、終了時には次ステップＳ６に進むが、未終了時にはステップＳ１に戻る。
【０４６３】
ステップＳ６において前ステップＳ２乃至Ｓ４で得たデータから対象者の体格について把握した後、次ステップＳ７で筋配置を認識する。つまり、対象者の体型はステップＳ４で得ているので、これによっておおよその筋配置は分かるが、ステップＳ２で得た対象者の外形形状やステップＳ３で得た内部構造データを使って、より精度の高い筋配置の認識及びモデルの高精度化を図ることができる。
【０４６４】
ステップＳ８では、上記干渉波シミュレーターによる方法を用いて効果器における使用電極を選択した後、次ステップＳ９で対象者に自らの意志により筋肉を動かしてもらう。その場合の動作パターンとしては、前腕の場合に、例えば、下記の動作例が挙げられる。
【０４６５】
・手を前に出して拳をつくった状態での、肘伸展、回内、第２乃至第５指のＤＩＰ、ＰＩＰ、ＭＰ屈曲、第１指のＩＰ、ＭＰ屈曲
・手で鉄アレイを把持して、これを持ち上げてからの回外、肘屈曲、指屈曲時の手首の掌屈
・手を前に出して掌面を平らにした状態における、対立（平坦状態）、第２乃至５指のＭＰ外転、第１指のＣＭ橈屈外転
・手首を下に曲げた状態からの指伸展時の手首の掌屈
・手首を上に曲げた状態からの手首の背屈
・手刀状態での、第２乃至第５指のＤＩＰ、ＰＩＰ、ＭＰ伸展とＭＰ内転
・親指を小指の付け根につけた状態での、対立（湾曲状態）、第１指のＣＭ尺屈内転。
【０４６６】
ステップＳ１０では、入力用セレクタ１０４ｉの制御によって使用電極を選択した上で筋電検出部２０１で筋電位情報を取得して周波数解析を行う。このときに取得したデータ（これを「マスターＥＭＧ」と定義する。）には、下記に示す周波数成分が含まれる。
【０４６７】
・筋トーヌスに係る成分
・運動単位に係る総和成分。
【０４６８】
尚、マスターＥＭＧについては、そのときの動作パターンを示すデータをリンクした上で所定の記憶手段に格納しておく。
【０４６９】
ステップＳ１１では、電気刺激のみによって対象筋を動作させ、次ステップＳ１２でそのときの筋電位情報を取得して周波数解析を行う。つまり、この場合に取得されるデータ（これを「スレーブＥＭＧ」と定義する。）には、下記に示す周波数成分が含まれる。
【０４７０】
・電気刺激信号の成分
・運動単位に係る総和成分。
【０４７１】
従って、スレーブＥＭＧから電気刺激信号の成分（これは電気刺激の原信号として既知であり、当該信号のＦＦＴ解析により直接算出できる。）を差し引くことで運動単位の総和成分が得られるので、これをマスターＥＭＧから除去すれば、筋トーヌスの成分だけを抽出することができる。これをその動作パターンに関連付けて記憶・学習すれば、その後のＥＭＧデータから筋の動作予測を容易に行える。つまり、電極から取得した筋電位情報から運動単位の特性を考慮することなく動作を予測できるようになる。
【０４７２】
例えば、腕立て伏せを１００回行う前と後でＥＭＧ測定を行った場合には、取得されるＥＭＧデータが異なるが、これは運動単位について筋肉疲労の影響が生じたためである。よって、従来の方法のようにＥＭＧデータから直接的に動作を予測しようとしても困難を極めることになるが、本発明によれば、ＥＭＧデータから運動単位に係る総和成分を除去した筋トーヌスの成分に基づいて動作予測が可能になる。
【０４７３】
しかして、上記した装置によれば、下記に示す利点が得られる。
【０４７４】
・従来のシャフトアーム等を使った外骨格型の機構を必要としないので、軽量化にとって有利であり、対象者はあたかも洋服を着用する感覚で使用することができる。
【０４７５】
・外力発生のための消費電力が少ない。つまり、サーボモータ等の動力源やエアシリンダの空圧源等が不要であり、対象者の筋肉への刺激付与によって関節の駆動や触覚呈示が可能である。
【０４７６】
・駆動時の雑音がなく、無音での力触覚呈示が可能になるので、例えば、ゲーム機への利用における効果音や環境音への悪影響がない。
【０４７７】
・ファントムセンセーションの利用により皮膚表面への擬似的な刺激を表現できる。
【０４７８】
・運動単位の特性や状態に影響されることなく、筋トーヌスの予測用データに基づいて筋の動作予測を行える。
【０４７９】
【発明の効果】
以上に記載したところから明らかなように、請求項１に係る発明によれば、対象者の筋肉に一定の電気刺激を与えながら、当該筋肉の筋電位情報を取得して周波数解析を行うことで筋細胞の収縮のみに起因する基礎データを予め求めておき、その後に筋電位情報の周波数解析により得られるデータから当該基礎データを除去したデータを算出することによって、実際に筋が動作する前に筋の動作を予測することができる。従って、筋電位情報から得られる純粋に筋トーヌスに係るデータに基づいて動作予測を行えるので、筋の疲労度等を含む各種の要因によって予測精度が影響を受け難くなる。
【０４８０】
請求項２に係る発明によれば、対象者がその意志で筋肉を動かそうとしたときに電極から検出される信号の周波数成分から、ＡＴＰによる多筋線維の動作集合状態信号の周波数成分を取り除いたデータに基づいて、筋の動作予測を行うことができる。
【０４８１】
請求項３に係る発明によれば、筋電位情報における運動単位の特性を考慮することなく筋トーヌスの予測用データを得ることができる。
【０４８２】
請求項４に係る発明によれば、対象者個人の筋特性に係るデータを取得することによって個性の認識が可能になる。
【０４８３】
請求項５に係る発明によれば、筋トーヌスの予測用データを対象者の動作パターンと関連付けて予め記憶・学習させておくことによって、動作パターンの予測を迅速に行うことができる。
【０４８４】
請求項６、１１、１６に係る発明によれば、電極を選択した実際に電気刺激を付与する前に筋肉の動作を予測することができるので、当該電気刺激によって対象者の意に沿わない動作が実行されたり、筋肉の動作において好ましくない刺激付与や動作変化が起きるのを未然に防止することができる。
【０４８５】
請求項７、１２に係る発明によれば、電極部と形状情報取得部とを同じシート状基材の上に積層して形成することで構成の簡単化を図ることができる。
【０４８６】
請求項８、１３に係る発明によれば、シート状基材の上に電極群を効率的に配置することができる。
【０４８７】
請求項９、１４に係る発明によれば、関節の動作によって巻き込まれることのない皮膚表面に対して電極群を無駄なく配置することができる。
【０４８８】
請求項１０、１５に係る発明によれば、効果器の人体への装着が容易になるとともに、装着時の蒸れを防止することができる。
【０４８９】
請求項１７に係る発明によれば、対偶毎にその形状に合った効果器を部分的に作成することができるので着用し易い。
【０４９０】
請求項１８に係る発明によれば、カメラ等の撮像手段を要することなく着用した効果器による圧力検出情報に基づいて対象者の外形形状を把握することができる。
【０４９１】
請求項１９に係る発明によれば、骨や筋等の内部構造を含むデータベースを得ることによりモデル精度の向上を図ることができる。
【０４９２】
請求項２０に係る発明によれば、色差の違いに基づく画像処理によって筋の識別を容易に行うことができ、筋の区別を自動化し易い。
【０４９３】
請求項２１に係る発明によれば、骨や筋等の形状を規定する点情報を容易に取得することで３次元形状データを作成することができる。
【０４９４】
請求項２２に係る発明によれば、３次元形状データの動的変化を容易に表現することができる。
【０４９５】
請求項２３に係る発明によれば、赤筋と白筋の占める量的な比率からを対象者の筋特性を把握することができる。
【０４９６】
請求項２４乃至２６に係る発明によれば、使用電極の選定にあたって干渉波のシミュレーションを事前に行うことで対象部位の刺激領域に適した電極位置を把握することができ、これによって実際の電気刺激付与時に理想的な干渉波刺激を得ることができる。
【０４９７】
請求項２７に係る発明によれば、電極候補の中から選定した電極の配置に応じて電気刺激の出力を設定することができ、低温火傷等の発生を防止できる。
【０４９８】
請求項２８に係る発明によれば、電極候補を事前に保存・登録しておくことで、実際の電気刺激付与時に電極候補から使用電極を直ちに選択することができる。
【０４９９】
請求項２９に係る発明によれば、電気刺激に付与にあたって最終的に選定された電極位置と同じ位置から筋電位情報を検出して周波数解析を行うことで筋トーヌスの予測精度を高めることができる。
【０５００】
請求項３０に係る発明によれば、対象者の関節についての動作予測を的確に行うことができる。
【０５０１】
請求項３１に係る発明によれば、同じ電極を使った電気刺激付与と筋電位情報の取得とが同時に競合しないように制御できる。
【図面の簡単な説明】
【図１】本発明に係る装置の基本構成を示す図である。
【図２】機能的電気刺激を利用した力触覚呈示装置の構成例を示す図である。
【図３】図４乃至図１４とともに効果器に使用する電極シートの構成について説明するための図であり、本図は電極シートを腕に巻き付ける様子を示す図である。
【図４】電極シートの分解斜視図である。
【図５】電極層３Ａを構成する最も皮膚側のシート４を部分的に示す図である。
【図６】電極層３Ａを構成する導通シート５を部分的に示す図である。
【図７】電極層３Ａを構成する配線シート６を部分的に示す図である。
【図８】電極層３Ａを構成する表面シート７を部分的に示す図である。
【図９】感圧・感温層３Ｂを構成する行電極シート８を部分的に示す図である。
【図１０】感圧・感温層３Ｂを構成する感圧・感温シート９を部分的に示す図である。
【図１１】感圧・感温層３Ｂを構成する列電極シート１０を部分的に示す図である。
【図１２】感圧・感温層３Ｂを構成する絶縁膜１１を部分的に示す図である。
【図１３】電極シートの一部を拡大して示す断面図である。
【図１４】腕に巻き付けた状態の電極シートを示す概略的な断面図である。
【図１５】図１６とともにデータグローブにおける指の屈曲状態を検出について説明するための図であり、本図はデータグローブの指部の背面を示す。
【図１６】指を曲げた状態を示す図である。
【図１７】抵抗パターン、圧力や温度の検出部を具備するデータグローブの要部構成を概略的に示す図である。
【図１８】人体の体型分類についての説明図である。
【図１９】基準体型モデルにおける諸量の定義を示す図である。
【図２０】体型形座標空間を構成する２次元座標平面についての説明図である。
【図２１】体型形座標空間におけるＺ軸に長さ比率の変数を設定した様子を示す図である。
【図２２】体型形座標空間におけるＺ軸に長さ比率及び重量比率の変数を設定した様子を示す図である。
【図２３】対象者から取得した画像データ及びその伸縮操作についての説明図である。
【図２４】図２５とともに対象者の３次元データについてのスライス処理に関する説明図であり、本図は立位姿勢の対象者（画像）とスライス間隔を示す。
【図２５】対象者の３次元データについてスライス処理と断層面との関係を示す図である。
【図２６】体型形座用空間を構成するＸ−Ｙ平面に設定した極座標系（ｒ，θ）において点の配置例を示す図である。
【図２７】多角形の重心計算によって対象者の体型形を特定する処理についての説明図である。
【図２８】重心Ｇ（Ｘｇ，Ｙｇ）からこれに対応する関数値を求める様子を示す図である。
【図２９】図３０とともに、身体に関する基本データの入力及びデータ加工に係る処理例を示すフローチャート図であり、本図は処理の前半部を示す。
【図３０】処理の後半部を示す。
【図３１】人体についての３次元データの取得及び運動に関するワイヤーフレームモデルの生成、運動に伴う重心位置データの取得に係る処理例を示すフローチャート図である。
【図３２】対象者の３次元データに関するスライス処理についての概略的な説明図である。
【図３３】人体ポリゴンＤＢとその構成ＤＢとの依存関係を示す図である。
【図３４】人体ポリゴンＤＢの生成に係る処理例を示すフローチャート図である。
【図３５】図３６乃至図４０とともに初期学習について説明するための図であり、本図は筒状に巻いた圧力検出用シートを示す。
【図３６】筒状に巻いた圧力検出用シートの別例を示す図である。
【図３７】圧力検出用シートについての素子選択について説明するための図である。
【図３８】圧力検出用シートに挿入して使用する基準円柱の組みを示す図である。
【図３９】初期学習の手順例を示すフローチャート図である。
【図４０】圧力検出用シートによる圧力分布情報から形状を取得する手順例を示すフローチャート図である。
【図４１】図４２乃至図４４とともに、上肢の形状認識について説明するためのものであり、本図は形状認識の手順例を示したフローチャート図である。
【図４２】対象者に視覚表示装置を装着して当該装置上に模倣すべき上肢の動きを映し出した状態を示す説明図である。
【図４３】人体の上肢を概略的に示す図である。
【図４４】圧力パターンの数例を示すグラフ図である。
【図４５】図４６及び図４７とともに断層面データから対象部位の内部構造を含む形状モデルを作成する処理例について説明するためのフローチャート図であり、本図は処理の始めの部分を示す。
【図４６】処理の中間部を示す。
【図４７】処理の終盤部を示す。
【図４８】各断層面とその面上の図形（三日月形状）について、輪郭線長に対するポリゴンポイント数の割り当てを示す説明図である。
【図４９】三日月形状及びその面重心Ｇと、当該重心Ｇを中心とする等角分割線によって輪郭線上にポリゴンポイントが生成される様子を示す説明図である。
【図５０】面重心位置の計算処理についての説明図である。
【図５１】面重心を中心とする等角分割線と輪郭線との交点として求まるポリゴンポイントを示す説明図である。
【図５２】刺激条件について諸量の定義を説明するための波形図である。
【図５３】図５４とともに干渉波シミュレーションについて説明するための図であり、本図は皮膚表面Ｓ上の１点と内部の対象部位ＸＰとの関係を示す概略図である。
【図５４】皮膚表面Ｓ上の３点と内部の対象部位ＸＰとの関係を示す概略図である。
【図５５】図５６とともに、ＡＴＰによる多筋線維の動作集合信号を取得する方法について説明するための図であり、本図は装置例を概略的に示す図である。
【図５６】ＦＦＴ後の周波数特性を概念的に示すグラフ図である。
【図５７】電極を通した電気刺激と、電極からの筋電位情報の取得処理とを時分割処理で行う場合の構成例を概念的に示す図である。
【図５８】図５９とともに人体の脂肪と刺激による主観評価法の一例を示すフローチャート図であり、本図はその前半部を示す。
【図５９】処理の後半部を示す図である。
【図６０】図６１乃至図７２とともに、本発明の実施例を示すものであり、本図は装置の全体構成を示すブロック図である。
【図６１】Ｉ／Ｏ処理部を主に示す図である。
【図６２】統括制御部の構成を示す図である。
【図６３】統括制御部とその周辺部との関係を示す図である。
【図６４】図６５とともに力触覚呈示装置としてゲーム機への適用例を示すものであり、本図は装置の全体構成を示すブロック図である。
【図６５】要部の構成だけを取り出して示す図である。
【図６６】図６７乃至図７１とともに、装置の制御方法について説明するためのフローチャート図であり、本図は制御の全体的な流れを示すフローチャート図である。
【図６７】基準面圧力情報の取得に関するフローチャート図である。
【図６８】関節角度に応じた面圧力情報の取得に関するフローチャート図である。
【図６９】図７０とともに、筋の仕事量による筋電情報の取得に関するフローチャート図であり、本図は処理の前半部を示す。
【図７０】処理の後半部を示す。
【図７１】電気刺激発生による力覚呈示の手順例を示すフローチャート図である。
【図７２】筋電位情報の取得と記憶・学習処理の一例を示すフローチャート図である。
【符号の説明】
１…モーションキャプチャー装置、１ａ…効果器、１ａｓ…形状情報取得部、１ａｄ…電極部、１ｂ…電気刺激発生部、１ｃ…形状認識部、１ｄ…制御手段、１ｆ…周波数解析部、１Ａ…力触覚呈示装置、１Ａｄ…効果器、２…電極[0001]
BACKGROUND OF THE INVENTION
  The present invention is a motion capture device that predicts muscle movement by detecting myoelectric potential information (in a virtual space created by a computer, a human image simulating an operator or an imaginary human image is displayed by moving image processing. An input device for acquiring the actual human body, the shape of a part of the body or the movement thereof as data suitable for computer processing)Force / tactile sensation presentation device that presents force and tactile sensation using electrical stimulation to musclesApplicable to.
[0003]
[Prior art]
  In fields such as virtual reality (virtual reality) and remote reality (telereality), power in a virtual space (or virtual world) created by a computer or a remote world (or remote environment, micro world, etc.) isolated from the operator 2. Description of the Related Art A force / tactile sensation presentation device using a force / tactile glove or the like is known as a man-machine interface for allowing a user to feel tactile sensation (force sensation and tactile sensation) and temperature sensation. This device is introduced to extend the virtual reality recognized mainly by vision and hearing to a more dynamic sense (force / tactile sensation), or to give a sense of reality to the operation sense in the remote world. For example, it acts on human hands and arms using electrical stimulation of muscles.
[0004]
In addition, use of "functional electrical stimulation" has been proposed for electrical stimulation to the skin surface and muscles. “Functional Electrical Stimulation” (FES for short) means an electrical stimulation method that assists or controls biological functions with a clear understanding of the purpose and understanding of the mechanism of action. It is mainly used to move a paralyzed limb by applying electrical stimulation to a target peripheral nerve or muscle via an electrode for a patient with motor paralysis.
[0005]
Briefly explaining the principle of FES, focusing on the fact that muscles contract by receiving motor movement commands from the brain as electrical signals, the potential is applied to motor neurons by electrical stimulation instead of electrical signals from the brain. It is intended to contract muscles.
[0006]
In FES, when controlling the contraction of a muscle, there are a case where an electrical stimulus is directly applied to the muscle and a case where an electrical stimulus is applied to a nerve bundle that controls the muscle. There is. One of them is called “centrifugal FES”, which stimulates motor nerve fibers in the nerve bundle to directly contract the muscles under its control to reconstruct the intended function. Is referred to as “afferent FES”, and by stimulating afferent nerve fibers in the same nerve bundle, the joint muscles of the muscles simultaneously contract and perform useful control by themselves.
[0007]
For example, when the common peroneal nerve is strongly stimulated, ankle dorsiflexion occurs at the same time as efferent FES, and at the same time a flexion reaction occurs in the hip and knee joints. This stimulates afferent fibers of the common peroneal nerve. This is because bending reflection occurs.
[0008]
As for the method of applying electrical stimulation to muscles, “surface electrode method”, “percutaneous electrode method”, and “implanted electrode method” are known from the difference in how electrodes are attached to a living body.
[0009]
[Problems to be solved by the invention]
By the way, in the conventional apparatus using EMG, it is possible to know the activity of the muscle by acquiring the electromyogram waveform from the electrode without stimulating the living body, but when the waveform is simply subjected to frequency analysis, Because only myoelectric potential information is obtained as data in which the frequency component due to muscle tonus and the frequency component accompanying the contraction of muscle cells are obtained (that is, in this case, muscle tonus appears and the muscle cell contracts) Myoelectric potential waveform is actually obtained data.) There is a problem that it is difficult to estimate the muscle movement before the muscle actually moves.
[0010]
Accordingly, an object of the present invention is to perform muscle motion prediction by calculating muscle tonus prediction data based on frequency analysis data of myoelectric potential information.
[0011]
[Means for Solving the Problems]
In order to solve the above-described problems, the motion capture device according to the present invention, the electrical stimulation device using the same, and the force / tactile sensation presentation device include the following basic components.
[0012]
An electrical stimulation generator for supplying electrical signals to a plurality of electrodes used in contact with the skin surface.
[0013]
-By calculating the difference data between the frequency component related to the electrical stimulation signal to the anode side electrode and the frequency component related to the detection signal from the cathode side electrode, the frequency component of the signal generated from the muscle by the electrical stimulation is calculated. Frequency analysis unit to analyze.
[0014]
The difference data related to the frequency component obtained by the frequency analysis unit is stored as data of the frequency component related to the motion aggregate state signal of the multi-muscle fiber by ATP, and the data is then detected from the electrode Control means for calculating a result of subtraction from a frequency component related to potential information and predicting muscle motion.
[0015]
Therefore, according to the present invention, basic data resulting only from muscle cell contraction is obtained in advance by acquiring myoelectric potential information of the muscle while performing constant electrical stimulation to the subject's muscle and performing frequency analysis. After that, by calculating the data obtained by removing the basic data from the data obtained by frequency analysis of myoelectric potential information (data for predicting muscle tonus), the muscle motion is predicted before the muscle actually moves can do.
[0016]
DETAILED DESCRIPTION OF THE INVENTION
  The motion capture device according to the present invention extracts in advance only the frequency component accompanying the contraction of myocytes based on the frequency analysis result of the myoelectric potential information acquired through the electrodes attached to the body surface of the subject, and thereafter The frequency component related to muscle tonus is identified from the myoelectric potential information obtained when the muscle is operated at the subject's will, and the muscle motion can be predicted immediately before that. The device is an electrical stimulation device with an effector for applying electrical stimulation to muscles,For example, force / tactile sensation display device for virtual reality and telerealityApplication to is possible.
[0017]
FIG. 1 shows a basic configuration of the apparatus, and the motion capture apparatus 1 includes an effector 1a, an electrical stimulus generation unit 1b, a shape recognition unit 1c, a control means 1d, and a frequency analysis unit 1f.
[0018]
The effector 1a includes an electrode portion 1ad composed of an electrode group that is used in contact with the skin surface. For example, the electrode portion 1ad is formed of a highly flexible electric insulating material as described later. A configuration in which a large number of electrodes are arranged on the formed sheet-like base material can be used.
[0019]
Moreover, the shape information acquisition part 1as acquires the information which shows the shape of a skin surface, and a shape change, As for the method of the information acquisition, the method by the optical method and pressure detection is mentioned.
[0020]
Examples of the former method include a method using a stereo (type) camera and a range finder, a method using a panning shot using an imaging means such as a CCD camera, and the like. For example, in the shape recognition method using two stereo cameras, a subject to be photographed is photographed (moving image or still image) with a reference camera and a detection camera arranged with different line-of-sight directions, and according to the parallax between the two cameras. A shape of an object can be calculated from a captured image, and an area-based matching method using an epipolar line is known. In the range finder, for example, slit light obtained by laser irradiation is reflected by a galvanometer mirror, and the reflection direction is gradually changed to perform light scanning on the subject, and the reflected light at that time is received by light receiving means ( For example, the shape information is acquired by reading with a silicon rangefinder detection chip or the like.
[0021]
In any case, it is necessary to obtain reflected light and image information from the target as primary information of the shape and perform shape recognition by subsequent image processing or the like, and it is necessary to install it as a device separate from the electrode unit. Therefore, a method using detection of pressure on the skin surface is preferable.
[0022]
That is, when it is desired to acquire data on a shape having a smooth curved surface such as the surface of the human body, for example, a shape or a shape change accompanying a bulge of a muscle on the surface of the human body, the following (1) to (3) The shape may be recognized by the procedure shown in FIG.
[0023]
(1) Prepare pressure detection means for detecting the pressure on the skin surface by wrapping or wearing the subject's body.
(2) The relationship between the detection information obtained by the pressure detection means and the surface shape is acquired in advance, and these are stored in the storage means as a data table or a function expression.
(3) Receiving detection information from the pressure detection means wound or worn around the subject, shape data is calculated based on the surface shape data corresponding to the information based on the information from the storage means.
[0024]
When detecting the pressure on the surface of the object in this way, acquiring the relationship between the pressure detection information and the surface shape in advance, storing and learning them as a data table or a function expression, and performing actual shape recognition The shape data can be obtained by referring to the stored and learned information. As will be described later, when a method for forming the pressure detecting means on the sheet-like base material in the same manner as the above-described electrode portion is used, the pressure detecting portion (a configuration example thereof will be described later) easily with respect to the electrode portion. There is an advantage that it can be attached. In addition to the pressure detection, it is useful to provide a temperature detection unit on the skin surface because, for example, a clue regarding the exercise state can be further obtained from the heat generation state of the muscle.
[0025]
The electrical stimulation generator 1b supplies electrical signals for low-frequency electrical stimulation and interference wave electrical stimulation to the electrode group constituting the electrode unit. The “low frequency electrical stimulation” is mainly used for surface muscles. For example, it is used to obtain a therapeutic effect by passing an electric current through the living body, and promotes fracture healing and wound healing, autonomic nervous system functions such as urination, respiratory functions such as respiratory pacing by phrenic nerve stimulation, heart Applications to circulatory system functions such as pacing are being made.
[0026]
Further, “interference wave electrical stimulation” is mainly used for stimulation of internal muscles or application of pseudo stimulation to the skin surface using phantom sensation. Note that “Phantom Sensation” means that a plurality of (for example, two) vibrators are installed at different locations on the skin, and the same frequency band is simultaneously applied to the vibrators, and the skin between the vibrators. If you feel the illusion that the center of the surface is vibrated, or if you change the frequency of another vibrator that is paired with respect to a certain vibrator, the position of the vibratory stimulus moves between the two vibrators. This is a phenomenon that gives you the feeling of being. In a force / tactile sensation presentation device to be described later, this is realized by electrical stimulation of a living body through an electrode group, and used to present a force sense to a subject.
[0027]
The electrical signal supplied to the electrode unit 1ad by the electrical stimulation generator 1b is preferably controlled separately for each muscle to be stimulated, and a plurality of channels of stimulation patterns can be set for each muscle. In addition, it is preferable to perform synchronization control for each channel.
[0028]
The shape recognition unit 1c is provided for recognizing the shape of the skin surface from the information obtained by the shape information acquisition unit 1as, and the recognition result is sent to the control means 1d. For example, the shape recognizing unit 1c receives detection information from the pressure detection means wound around or attached to the subject and calculates surface shape data corresponding to the information. In other words, a numerical relationship between the surface pressure P and the shape S (position change in the normal direction of the surface) (which is expressed by a functional expression “S = F (P)”) is a free curve (spline curve). If the information necessary for specifying the free curve is stored in the storage means in advance, the surface pressure at a specific position is P = Pa. In addition, the shape data can be calculated as “S = Sa = F (Pa)”. Needless to say, a known reference method or an interpolation method such as a method of creating a database of function relationships as a data table instead of a function formula and referring to data after pressure detection can be used.
[0029]
The frequency analysis unit 1f is the difference data between the frequency component related to the output signal of the electrical stimulation to the anode side electrode and the frequency component of the signal detected from the cathode side electrode among the electrodes constituting the electrode unit 1ad. It has a role of calculating (level difference), and analyzes (Fourier analysis) a frequency component of a signal generated from a muscle by electrical stimulation. That is, it has a role to extract frequency information of myoelectric potential due to muscle cell contraction caused by electrical stimulation, and the analysis result is sent to the control means 1d.
[0030]
The control means 1d realized by a computer such as a computer constructs a database (to be described later) showing the arrangement of bones, muscles, fat, and skin of the human body related to the subject, as well as information on the muscle arrangement and the shape. After selecting an electrode to which the electrical stimulation signal is to be supplied from the electrode group based on the recognition result of the shape of the muscle and the shape change from the recognition unit 1c, the electrical stimulation is generated so as to supply the electrical stimulation signal to the electrode. A control signal is sent to the unit 1b. In addition, the control unit 1d stores the difference data related to the frequency component acquired by the frequency analysis unit 1f as data of the frequency component related to the action set state signal of the multimuscle fiber by ATP (adenosine triphosphate), Thereafter, a result obtained by subtracting the data from the frequency component (analysis result by the frequency analysis unit 1f) related to the myoelectric potential information detected by the electrode of the electrode unit 1ad is calculated to perform muscle motion prediction.
[0031]
In FIG. 1, the control unit 1d includes a control unit 1ds and an input / output (I / O) control unit 1dt, and the input / output (I / O) selection processing unit 1e sends the data to the input / output (I / O) selection processing unit 1e. The following switching is performed according to the signal to be transmitted.
[0032]
Selection of each electrode constituting the electrode unit 1ad and alternative selection between the electrode unit 1a and the d shape information acquisition unit 1as
Selection of shape recognition unit 1c or frequency analysis unit 1f
[0033]
When performing such selection control by time-sharing processing, for example, when applying electrical stimulation, the output terminal of the electrical stimulation generating unit 1b is connected to the electrode set as the anode among the electrodes in the electrode unit 1ad. (In actuality, an electrode selected from a large number of electrodes is set as an anode electrode, but the figure implicitly indicates that the electrode is switched by the symbol of a single switch.) Of these, the electrode set as the cathode is selected in the switching unit sw1 (indicated by the symbol of the switch in the figure), and this electrode and the frequency analysis unit selected by the switching unit sw2 (indicated by the symbol of the switch in the figure) The 1f input terminal is connected. As a result, the electrical stimulation generator 1b receives a control command from the controller 1ds to generate an electrical stimulation signal, and the signal is supplied to the anode electrode in the electrode unit 1ad via the input / output selection processing unit 1e (FIG. (A signal line to the electrode group is typically represented by a single line.) Then, the myoelectric potential information at that time is sent from the cathode electrode to the frequency analysis unit 1f through the input / output selection processing unit 1e, and the analysis result is sent to the control unit 1ds.
[0034]
When the shape information acquisition unit 1as and the shape recognition unit 1c are selected in the switching units sw1 and sw2, the information obtained by the shape information acquisition unit 1as is recognized through the input / output selection processing unit 1e. The signal is sent to the unit 1c (the signal lines are representatively shown as a single line in the figure), and the shape recognition result is sent to the control unit 1ds.
[0035]
If it is desired to acquire the myoelectric potential information and the shape information without passing through the input / output selection processing unit 1e, the shape information acquisition unit 1as and the shape recognition are performed between the frequency analysis unit 1f and the electrode unit 1ad. What is necessary is just to connect directly between the parts 1c (Thereby, two types of information can be acquired simultaneously).
Thus, the motion capture device 1 can be applied to the electrical stimulation device as it is, as can be seen from the configuration including the electrical stimulation generation unit 1b (in FIG. 1, the electrical stimulation generation unit 1b). Therefore, by predicting the movement of the muscle before actually applying the electrical stimulation to the subject's muscle, the state of the muscle can be adjusted so that the electrical stimulation is always performed properly. Can be monitored. For example, it is effective in terms of safety management because it can prevent muscle pain and damage in the case where a muscle motion unsuitable for the subject is caused by electrical stimulation.
[0036]
The configuration shown in FIG. 1 shows a basic configuration for predicting muscle movement, and using this, other controlled objects such as manipulators and artificial hands, or power assist devices (power assist devices) can be used. In order to control the operation and the state, as shown in FIG. 1 surrounded by a two-dot chain rectangle, a control signal (for example, the motion of the subject acquired by the motion capture process) is transmitted from the control unit 1ds to the drive control mechanism 1g. Command signal for causing the controlled object to perform the same movement as in FIG. 1, and the controlled object 1h (the figure shows a manipulator created by imitating the structure related to the degree of freedom of the human hand and arm) .) Can actually be driven. That is, the myoelectric potential information is acquired and the movement of the subject is recognized, and the prediction of the muscle tonus makes it possible to drive the controlled object 1h without a significant delay time. It is possible to provide a man-machine interface that is faithfully adapted to the sensation and allows fine control.
[0037]
FIG. 2 shows an application example to a force / tactile sensation presentation apparatus. The force / tactile sense presentation apparatus 1A includes an input unit 1Aa, a control unit 1Ab (including the above-described electrical stimulation generation unit 1b), and an input / output buffer group 1Ac. (Input buffer group 1Aci and output buffer group 1Aco), an effector 1Ad, a model construction / calculation processing unit 1Ae, an image processing unit 1Af, a display unit 1Ag, a pressure / temperature recognition unit 1Ah, and a pressure / temperature detection unit 1Ai. .
[0038]
The input unit 1Aa includes all input means such as a keyboard and a pointing device, and is necessary for manual or automatic input of data. Information acquired thereby is obtained from the control unit 1Ab and the model construction / calculation processing unit. 1Ae.
[0039]
The control unit 1Ab sends an electrical signal to the effector 1Ad via the output buffer group 1Aco located at the subsequent stage, or myoelectric potential information (or electromyogram information) from the effector 1Ad via the input buffer group 1Aci. ). The output terminal of each buffer constituting the output buffer group 1Aco and the input terminal of each buffer constituting the input buffer group 1Aci are respectively connected to each electrode (described later) of the electrode array provided in the effector 1Ad. It is connected. In addition, the pressure information and temperature information on the skin surface detected by the pressure / temperature detection unit 1Ai attached to the effector 1Ad are sent to the control unit 1Ab via the pressure / temperature recognition unit 1Ah. .
[0040]
Since the effector 1Ad is used in contact with the skin of the human body, the effector 1Ad is formed with a shape that matches the shape of the human limb, part of the body, and the like. For example, the form includes a glove, a body suit (or wet suit), an arm band, a finger sack, a sock, and the like. FIG. 2 shows an effector attached to the human body and used on the arm. Moreover, in order to present a sense of temperature in addition to force and touch, a temperature range in which a large number of thermal control elements (such as Peltier elements) are incorporated into the effector 1Ad and no danger to the human body is achieved by thermal control of the elements. It is more effective if it is configured so that it can feel heat and temperature.
[0041]
As for the electrical stimulation method for a living body, the surface electrode method is adopted in the present invention with respect to how to attach the electrodes. The reason is that the method can be used without damaging the human body, and that the percutaneous electrode method and the embedded electrode method have problems such as electrode and wiring breakage and infection.
[0042]
FIG. 3 to FIG. 14 illustrate examples of effector configurations suitable for this surface electrode method, and show an example of an electrode sheet 3 having a large number of electrodes 2, 2,... (Electrode array). .
[0043]
The electrode sheet 3 is used by being attached to the skin surface of a human body. For example, as shown in a simplified manner in FIG. In this example, the electrode sheet 3 is formed in a rectangular shape, the length of one side is substantially the same as the length of the forearm A, and the length of the other side is substantially the same as the circumferential length of the forearm A. It is the same. As a result, when the electrode sheet 3 is wound around the forearm A, the forearm A is almost covered with the electrode sheet 3 on its peripheral surface.
[0044]
FIG. 4 schematically shows the laminated structure of the electrode sheet 3. A four-layer pressure-sensitive / temperature-sensitive layer 3B is formed on a four-layer electrode layer 3A, and an insulating layer is further formed thereon. It has an example in which 3C is laminated.
[0045]
3 A of electrode layers have the structure which laminated | stacked the sheet | seat 4, the conduction | electrical_connection sheet 5, the wiring sheet 6, and the surface sheet 7, as shown on the left side of the figure.
[0046]
The sheet 4 is a sheet positioned on the most skin side, and there are a large number of small electrode holes 4a, 4a,... (See FIG. 5) having a polygonal shape (for example, a square or a regular hexagon) in a matrix shape. Are formed, and the electrodes 2, 2,... Are regularly formed on the sheet by filling the electrode holes 4a, 4a,. Each electrode hole 4a is formed with a ridge portion 4b surrounding the periphery thereof. The ridge portion 4b has a substantially trapezoidal cross-sectional shape in the thickness direction, and corresponds to a short side of the trapezoid. The end portion is a contact portion with the skin, and the portion corresponding to the hypotenuse forms a tapered surface over the periphery of the electrode hole 4a. In addition, this protrusion part 4b is provided in order to make the adhesiveness of the electrode 2 and skin favorable (refer FIG. 5, FIG. 13).
[0047]
The conductive sheet 5 is laminated on the surface side of the sheet 4, and in the conductive sheet 5 at positions corresponding to the electrodes 2, 2,. 5a, 5a,... (See FIG. 6) are formed, and these energization holes 5a are filled with a conductive polymer, which are defined as energization portions 5b, 5b,. The conductive sheet 5 is provided in order to prevent short circuit between wiring patterns described later formed on the wiring sheet 6 and electrodes 2, 2,.
[0048]
The wiring sheet 6 is laminated on the surface side of the conductive sheet 5 (see FIG. 7), and terminal portions 6a, 6a,... And wiring patterns 6b, 6b,. That is, the wiring sheet 6 is formed with terminal holes having the same size as the current-carrying parts 5b, 5b,..., And filled with the conductive polymer so that the terminal parts 6a, In addition, wiring patterns 6b, 6b,... Extending in the longitudinal direction of the wiring sheet 6 are formed for the terminal portions 6a, 6a,. These wiring patterns 6b, 6b,... Are made of a conductive polymer, and are arranged so that they do not contact each other (it is difficult to ensure a sufficient wiring formation area on the sheet). In some cases, it is desirable to use a multilayer structure.) And the edge part on the opposite side to terminal part 6a, 6a, ... among each wiring pattern 6b, 6b, ... extends to the edge vicinity of the wiring sheet 6, and these are the terminal parts of an electrical stimulation generation | occurrence | production part. It is connected to (input / output terminals of the input / output buffer group 1Ac).
[0049]
The top sheet 7 (see FIG. 8) includes an (electrical) insulating sheet having the same size as the other sheets 4 to 6, for example, a synthetic resin material such as an insulating polymer, nylon, polypropylene (PP), or silicone. Is used and is attached to the wiring sheet 6.
[0050]
In the electrode layer 3A having such a structure, each electrode 2, the energizing portion 5b, the terminal portion 6a, the wiring pattern 6b, and the like can be formed by silk printing, for example.
[0051]
Next, formation of the pressure-sensitive / temperature-sensitive layer 3B will be described (see FIGS. 9 to 12).
[0052]
As shown in FIG. 4, the pressure-sensitive / temperature-sensitive layer 3 </ b> B has a configuration in which the row electrode sheet 8, the pressure-sensitive / temperature-sensitive sheet 9, the column electrode sheet 10, and the insulating film 11 are stacked from the side closer to the skin. .
[0053]
The row electrode sheet 8 (see FIG. 9) is formed by placing parallel electrodes 8a, 8a,... Made of conductive polymer in a predetermined direction on an electrical insulating film formed on a film such as polyester. The electrodes are formed along the longitudinal direction, and each electrode is insulated and separated by interposing an electric insulating film between these electrodes.
[0054]
The pressure-sensitive / temperature-sensitive sheet 9 (see FIG. 10) has a configuration in which pressure-sensitive / temperature-sensitive portions 9a, 9a,... Having a polygonal shape (square in the figure) are arranged at regular intervals. These correspond to the electrodes 2, 2,... In a one-to-one correspondence with each other. In addition, about the material used for the pressure sensitive / temperature sensitive part 9a, in the case of a pressure sensitive material, the pressure sensitive conductive rubber and piezoelectric material which have a stretching property are mentioned, for example. Examples of the conductive rubber include a pressure-sensitive conductive rubber in which a conductive substance such as metal powder or carbon fiber is dispersed in the rubber. As the piezoelectric material, a piezoelectric polymer material (such as vinylidene fluoride) is used. And a dielectric constant polymer. In the case of a heat-sensitive (temperature) material, for example, a heat-reactive material (thermistor constituent material) made by mixing manganese (Mn), nickel (Ni), or the like is used. Needless to say, the pressure-sensitive and temperature-sensitive parts 9a are isolated from each other by an insulating material.
[0055]
The column electrode sheet 10 (see FIG. 11) has the same structure as the row electrode sheet 8 except that the formation direction of the parallel electrodes 10a, 10a,. That is, the formation direction of these electrodes 10a made of a conductive polymer is defined so as to be orthogonal to the formation direction of the parallel electrodes 8a of the row electrode sheet 8 with the pressure-sensitive / temperature-sensitive sheet 9 interposed therebetween. The pressure sensitive / temperature sensitive parts 9a, 9a,... Are arranged separately at the intersections of the two electrode groups. As a result, a matrix arrangement (or a lattice arrangement) is established in which the electrode groups are orthogonal to each other. When integer variables “i” and “j” are introduced as indices, the i-th row and j-th column are designated. The pressure can be detected near the corresponding intersection position.
[0056]
The insulating film 11 (see FIG. 12) is formed using an electrically insulating material, for example, insulating silicone, polymer gel, etc., and is formed on the outermost surface of the pressure-sensitive / temperature-sensitive layer 3B to form a column electrode. The sheet 10 is insulated.
[0057]
The insulating layer 3C (see FIG. 4) is located on the most surface when it is attached to the forearm A, and a rubber material or the like is used for this.
[0058]
Thus, when such an electrode sheet 3 is wrapped around the forearm A and given stimulation by interference waves, the electrodes 2, 2,... Of the sheet 4 are applied to the skin of the forearm A (skin 13 and FIG. 14), when a predetermined frequency current is supplied to the pair of electrodes 2, 2,... From the control unit 1Ab through the output buffer corresponding to the electrodes, the forearm An interference wave is generated inside A, and the part of muscle M where the waves are strengthened by the interference can be stimulated. For example, if a frequency current of 4000 Hz (hertz) and a frequency current of 4010 Hz are supplied to the electrodes 2 and 2, interference occurs near the center on the line connecting the electrodes, and the muscle M located at the interference site It contracts when stimulated at a frequency of about 10 Hz. In addition, when performing low-frequency stimulation on the surface muscles on the skin surface, using two electrodes as the electrodes, after specifying each position of the pair of electrodes, a pulse waveform of about 100 Hz or less, for example, about 10 to 60 Hz ( A stimulus may be applied using a negative rectangular wave or the like.
[0059]
In detecting the pressure distribution on the skin surface, desired electrodes of the row electrode sheet 8 and the column electrode sheet 10, that is, electrodes corresponding to the i-th row and j-th column for the integer variables i and j (row electrode sheet 8). If the pressure sensitive material is pressed according to the pressure in the vicinity of the intersection of the electrodes when the i-th electrode and the j-th electrode of the column electrode sheet 10 are selected, the change in the capacitance is detected. be able to. That is, a capacitor including a pressure-sensitive / temperature-sensitive portion 9a is formed between the i-th electrode in the row electrode sheet 8 and the j-th electrode in the column electrode sheet 10, and the equivalent capacitance value C is determined at that location. Therefore, it is possible to obtain information related to the pressure P by back-calculating from the equivalent capacitance value C (that is, a desired position by matrix processing on the detection data in the i-th row and the j-th column). Pressure distribution data specifying the local pressure and the target range at can be obtained.) The control unit 1Ab obtains such pressure information from the pressure recognition unit 1h.
[0060]
Further, in detecting the temperature distribution on the skin surface, when a desired electrode of the row electrode sheet and the column electrode sheet, that is, the integer variables i and j are introduced as an index, it corresponds to the i-th row and the j-th column. By selecting the electrodes (the i-th electrode of the row electrode sheet and the j-th electrode of the column electrode sheet), it is possible to detect a change in the electrical resistance value of the thermosensitive material according to the temperature in the vicinity of the intersection of the electrodes. . That is, the resistance value R between the i-th electrode in the row electrode sheet and the j-th electrode in the column electrode sheet is a function of the temperature T at that location. Such information can be obtained. In this way, temperature distribution data specifying a local temperature and a target range at a desired position can be acquired by matrix processing on the detection data in the i-th row and j-th column (for example, 1 ≦ i ≦ N If 1 ≦ j ≦ M (N and M are natural numbers), a matrix of N rows and M columns is obtained as temperature data, and a specific range is selected and average temperature, variance, deviation, etc. are calculated. Can do that.)
[0061]
The pressure-sensitive / temperature-sensitive sheet 9 has a multilayer structure, that is, a pressure-sensitive sheet (the pressure-sensitive / temperature-sensitive portion 9a is formed only of a pressure-sensitive material between a row electrode sheet and a column electrode sheet. ) And a structure in which a temperature-sensitive sheet is sandwiched between the row electrode column and the column electrode sheet (the pressure-sensitive / temperature-sensitive portion 9a is formed of only a heat-sensitive material). A multilayered configuration may be adopted.
[0062]
As described above, when the electrode part and the shape information acquisition part are formed on the sheet-like base material with a multilayer structure (for example, a configuration in which a pressure or temperature detection layer is laminated on the electrode layer). When the effector is made into a shape that can be worn and used on the human body (for example, adopting a wet suit shape), it is easy to put on and take off, so that it takes just enough preparation time to change clothes. There are advantages.
[0063]
A so-called data glove is used in order to acquire data on the shape or motion of the hand or finger by wearing it on the human hand.
[0064]
15 to 17 are explanatory diagrams of the detection principle in the data glove, and the detection method is as follows.
[0065]
(I) Detection method for bent state of hand or finger = Method for grasping by change in resistance value accompanying change in length of resistance pattern (or resistor pattern)
(II) Detection method of temperature and pressure of hand and finger = Resistance change of heat sensitive material is used for temperature detection, and change of capacitance of pressure sensitive material is used for pressure detection. Method.
[0066]
FIG. 15 shows a configuration example in which a resistance pattern RP is formed on a glove base material G (formed of an insulating material) in order to detect a bent state of a finger portion. The resistance pattern RP is formed using a material (carbon or the like) whose resistance value changes according to the amount of elongation.
[0067]
The U-shaped resistance pattern RP extends along the longitudinal direction on the back surface of the finger, “L” indicates a natural length, and “R” indicates a resistance value at that time.
[0068]
As shown in FIG. 16, when a finger is bent with a glove on the hand, the resistance pattern RP is stretched by bending the finger (the amount of elongation is “ΔL” (> 0)), thereby causing resistance. The value R changes to “R + ΔR” (ΔR ≠ 0). That is, in this example, a DIP joint (Distal Palangial joint: the joint located at the most distal end in the second to fifth fingers) or a PIP joint (Proximal Palangial joint: proximal interphalangeal joint). That is, the resistance value change ΔR is caused by the change in the length of the resistance pattern RP (see the small circle in the figure) corresponding to the bending of the second to fifth fingers (the joint located second from the tip). Become.
[0069]
Thus, in the above (I), a resistance pattern for detecting the bending state of the hand part or the finger part as a change in resistance value is formed on the base material of the data glove formed of an electrically insulating material, When the joint of the hand or finger is bent, the length of the resistance pattern changes, and at this time, the resistance value of the resistance pattern changes to detect the bending state of the joint. it can. In addition, when a material in which the resistance pattern length “L” does not change according to the bent state (ΔL≈0) and only the resistance value at this time changes is used, Since the influence of stress due to repetition becomes a problem as a change with time, it is desirable to use a material whose resistance value changes as the length of the resistance pattern changes as described above.
[0070]
For (II), the same method as described for the effector described above may be adopted, whereby the temperature detection and pressure detection have the same structure (however, the materials used for detection are different). ) Can achieve a reduction in thickness and cost.
[0071]
FIG. 17 schematically shows the main structure of a data glove having a resistance pattern, a pressure and temperature detector, and a base material (not shown) formed of an insulating material (considering the paper surface in the figure as a base material). May have a U-shaped resistance pattern RP (shown with diagonal lines), pressure and temperature detection element portions DP, DP,... .) Are stacked in the thickness direction. As described above, the detection element portion DP is a heat-sensitive material interposed between the row electrode and the column electrode (these electrodes are not in contact with the resistance pattern RP or with each other). Alternatively, it may be composed of a pressure-sensitive material, and a structure in which a heat-sensitive layer and a pressure-sensitive layer are laminated may be used, and each detection element portion DP may be a pressure detection element or a temperature detection element. Can be defined as a planar pattern arrangement (for example, pressure detection elements and temperature detection elements are alternately arranged).
[0072]
In FIG. 17, a configuration in which a part of the electrode pattern for pressure or temperature detection and a part RP1 of the resistance pattern RP are shared (that is, a configuration in which the electrode pattern is also used as a resistance pattern) is adopted. The number of wirings can be reduced.
[0073]
Next, the model construction / calculation processing unit 1Ae shown in FIG. 2 will be described. This is realized by a computer (processing) device constituted by a computer, a memory, or a predetermined recording medium, and a database for constructing a numerical model related to a human body structure (hereinafter, “database” is abbreviated as “DB”). Is to create, maintain and manage it. All peripheral devices used to acquire basic data necessary for constructing a human body numerical model are included in the input unit 1Aa.
[0074]
The image processing unit 1Af exchanges information with the control unit 1Ab and the model construction / calculation processing unit 1Ae, sends a video signal to the display unit 1Ag, and displays the image. For example, image information is displayed to convey an object (virtual object) in a virtual space that appears as a CG (computer graphics) display as visual information to the user of the apparatus, or a numerical model of a human body is constructed. Perform screen display necessary for work. In the case of handling auditory information in addition to visual information, an audio signal processing unit may be included in the image processing unit 1Af and the audio information may be transmitted to the apparatus user through audio output means such as a speaker. When information printing is required, printing means such as a printer is attached to the image processing unit 1Af.
[0075]
Regarding the method of constructing the numerical model of the mechanical structure related to the human body, the method already proposed by the present applicant in Japanese Patent Application No. 10-266 can be followed.
[0076]
To briefly explain the point, the numerical model includes, for example, the following models.
[0077]
(I) Skeletal model
(Ii) Muscle model
(Iii) Nerve (motor nerve, sensory nerve, etc.) model
(Iv) Skin model
(V) Fat and internal organ model
The nerve model of (iii) does not include cranial nerves that perform advanced information processing, but includes nerves that are directly related to the mechanical structure and movement of the object, and (iv) the skin model includes In addition to skin surface condition and movement, wrinkles and electrical properties such as insulation resistance, the amount of wrinkles due to joint movement etc. is treated by replacing it with skin aging. (E.g. how to apply different skins and body hairs to each skin area covering the joints based on the angle, distance from the joints, and muscle tension, etc.) It is preferable from the viewpoint of reduction (because the method of calculating the approach of the heel by the simulation calculation based on the state of the joints and muscles has the disadvantage that the calculation is complicated and the calculation amount is too large). Regarding the model of fat and viscera in (v), attention is paid to the influence of the weight and position on the balance of the body and the movement of the center of gravity, the electrical resistance, etc., not the specific structure related to the function.
[0078]
Hereinafter, (i) and (ii) which are particularly related in the present invention will be mainly described.
[0079]
The simplest numerical model of the human body structure includes only the skeleton model (i), which can be created by the following procedure.
[0080]
(1) The body type is classified, and skeleton data including all bone shapes, lengths and weights is prepared for each classified standard body type.
(2) Input height, weight, and outline data related to the subject
(3) Identify the subject's body shape from the input data of (2), perform interpolation calculation based on the skeleton data of each reference body shape, and calculate the conversion ratio of the skeleton data related to the subject
(4) Based on the conversion ratio of (3) and the height and weight of the subject, the length and weight of each bone of the subject are determined to create a numerical model of the skeleton.
[0081]
First, in the step (1), the human body reference body type is classified into, for example, a lean type, a warrior type, and an obese type as shown in FIG. That is, the slime type (asthenicus; hereinafter abbreviated as “as”) is a thin body type, and the warrior type (atheticus; hereinafter abbreviated as “at”) is a body shape in which the chest torso is an inverted triangle. The obese type (piknices, hereinafter abbreviated as “pi”) is an enlarged body type such as the abdomen.
[0082]
Then, select one human body representing each body type and acquire skeleton data (including bone shape, length, weight) for each subject, or existing data from human body textbooks, etc. Use. That is, a numerical model (hereinafter referred to as “reference body type model”) is created by examining all information relating to the skeleton of each human body representing the lean type, the warrior type, and the obese type and creating a database. In this case, if an influence on the reference body shape is recognized due to a difference in sex, it is preferable to create a model in consideration of this. To do so, prepare a different reference body model for each gender, or prepare a reference body model for each body for one of the two genders, and for the other gender, use the conversion ratio for that reference body model. There is a method of deriving a reference body model of each body type from the data shown. For each bone shape, it is preferable to use a format suitable for image display (three-dimensional display or the like) on a computer as three-dimensional model data (for example, polygon data or the like).
[0083]
For the acquired skeletal data, in addition to the length and weight values of each bone, data on the ratio (ratio) to the height and weight is obtained, and the subject is based on these ratio data as described later. Used when calculating ratio data.
[0084]
Further, since the comparison of data is troublesome if the height of the human body representing each body type is different, it is preferable to prepare data when a reference height is set and each data is converted into the height.
[0085]
In the model M shown in FIG. 19, “HEIGHT_ref” indicates the height (reference height) of the reference body model, and the variable “HEIGHT_xx_XX” indicates the vertical length of each part of the human body in the standing posture (“xx "Indicates a part of the human body, and" XX "indicates a reference body shape.) For example, “HEIGHT_arm_u_as” indicates the length of the upper arm in the lean reference body model, and “HEIGHT_chest_at” indicates the chest length in the warrior reference body model.
[0086]
“HEIGHT” in this case indicates the length of each part of the human body including the skeleton and muscles. If “HEIGHT” is the length of the bone in the vertical direction or longitudinal direction, “HEIGHT_xx_XX” is the reference height “HEIGHT_ref”. The bone length ratio can be calculated from the divided ratio (hereinafter referred to as “_hr_xx_XX”).
[0087]
Similarly, by dividing the variable “WEIGHT_xx_XX” in which “HEIGHT” is replaced with “WEIGHT” indicating the weight of a human body part and each bone by the weight “WEIGHT_ref_XX” (hereinafter, “_mr_xx_XX”) You can ask.)
[0088]
FIG. 20 is a graph showing the concept of the above-described body shape on a two-dimensional coordinate plane (XY plane). Three axes “Ax_XX” (XX = as, at, pi) starting from the origin O are shown. Are set to have an angular interval of 120 ° with respect to each other.
[0089]
And the intersection P_XX of the circle cir centered on the origin O and each axis Ax_XX shows the position occupied by each reference body model. That is, the point “P_as” indicates the position on the XY plane for the lean model, the point “P_at” indicates the position on the XY plane for the warrior model, and the point “P_pi” is obese The position on the XY plane for the model of the mold is shown.
[0090]
For example, in the case where the subject has a body type that is located exactly between the lean type and the warrior type, with respect to two axes Ax_as (in the figure, the axis that coincides with the positive axis of Y) and Ax_at through the origin O Each point on the axis Bx extending at an angle of 60 ° (for example, the intersection point Q between the axis Bx and the circle ci) indicates the body shape of the subject.
[0091]
The meaning of each axis when a polar coordinate system composed of the r axis and the θ axis is set with respect to the XY plane will be described in detail later.
[0092]
FIG. 21 shows a space (hereinafter referred to as “body shape coordinate space”) in which an orthogonal axis (Z axis) is given to the XY plane of FIG.
[0093]
For example, when the above-described length ratio “_hr_xx_XX” is taken as the Z-axis, the point “H_xx_XX” indicating the value of “_hr_xx_XX” corresponds to each point P_XX.
[0094]
Now, if the part indicated by xx is the humerus (humerus) and the body type indicated by XX is the lean type (as), the value indicated by the point “H_humerus_as” (that is, the perpendicular line hanging from the point to the XY plane) The height of the foot (Z value) indicates the ratio of the length of the humerus to the reference length (height) in the lean reference body model.
[0095]
In order to calculate the Z value corresponding to the point Pt (xt, yt) on the XY plane, the following procedure is adopted.
[0096]
First, when the point “H_xx_XX” for each of the three points P_XX (XX = as, at, pi) is determined, one plane (πh_xx) passing through these points can be determined. That is, by selecting any two of the three points “H_xx_XX”, two vectors connecting the two points can be created (for example, a vector from the point “H_xx_pi” to the point “H_xx_as” and a point “H_xx_pi”). ”To the point“ H_xx_at ”, etc.), so that the normal vector in the direction orthogonal to both vectors is the vector n (a, b, c) (where a, b, c are the X, Y, and Z axes, respectively) The plane πh_xx can be expressed by the equation “a · X + b · Y + c · Z = d” (where d is a constant). Therefore, for example, if the point Pt (xt, yt) on the XY plane is determined, the Z value can be obtained by substituting X = xt and Y = yt into the above equation.
[0097]
Note that there are a large number of planes πh_xx for each part indicated by xx. In this sense, the Z axis can be regarded as a variable axis that collectively represents multiple variables as one axis.
[0098]
FIG. 22 shows the situation when the weight ratio “_mr_xx_XX” is further added to the Z axis of the body shape coordinate space, and the point “M_xx_XX” indicating the value of “_mr_xx_XX” corresponds to each point P_XX. To do.
[0099]
For example, if the part indicated by xx is the humerus and the body type indicated by XX is the lean type (as), the value indicated by the point “M_humerus_as” (that is, the perpendicular line hanging from the point to the XY plane) The height of the foot (Z value) indicates the ratio of the humerus to the body weight of the weight in the lean reference body model.
[0100]
The same method as described above is used to obtain a Z value corresponding to a point Pt (xt, yt) on the XY plane.
[0101]
That is, when each point “M_xx_XX” for each of the three points P_XX is determined, one plane (πm_xx) passing through these points can be determined (the normal vector of the surface is a vector nn (aa, bb, cc)) When the point Pt (xt, yt) on the XY plane is determined, the formula “aa · X + bb · Y + cc · Z = dd” (where dd is a constant)) By substituting xt and Y = yt into the above equation, the corresponding Z value can be obtained.
[0102]
In FIG. 21 and FIG. 22, there are only three reference body models, so in general, the function expression “Z = Fn (X, Y)” (where n is the above-described _hr_xx or _mr_xx). The curved surface represented by (1) is a plane (that is, expressed by a linear expression of X, Y, and Z), but the number of reference body models is increased, or a large number of subjects The accuracy of the function formula (surface expression) is improved by performing interpolation processing (for example, Bezier, spline interpolation, etc.) using the accumulation result of the data (body shape, ratio, etc.) Of course you can.
[0103]
In determining the reference body model, there are methods for selecting a plurality of human bodies belonging to each body type, or performing data averaging processing, etc., but in the following, simplification of explanation and priority of understanding will be given priority. Therefore, it is assumed that there is one human body representing each reference body type, and that the sex is fixed to one of the two sexes and that the influence of age is ignored.
[0104]
In the next step (2), height, weight, and outer shape data are input as basic data related to the subject.
[0105]
As an input method, there are a method of inputting manually and a method of acquiring an input value from image data of a subject.
[0106]
For example, with regard to height and weight, there are a method of directly inputting numerical values using an input means such as a keyboard when the values are known by the subject's knowledge, and a method of automatically inputting values measured by the height and weight scale. is there.
[0107]
As for the shape, the image data and three-dimensional data of the subject person are acquired (for example, using a stereo shooting camera, a hologram camera, a range finder, etc.) and the shape of each part of the human body is recognized. A method for automatically inputting data is mentioned. For example, a method using light interference fringes (shooting a subject by generating interference fringes of monochromatic light or three primary colors on the subject, and determining the depth of the subject in the photographing direction from the formation interval of the interference fringes generated on the subject ( The procedure for obtaining the three-dimensional data by obtaining the data on the projections and depressions and attaching the subject image without interference fringes to the data is briefly shown as follows.
[0108]
(1) Wearing markers for detecting the center of gravity (subjects for image processing, radio wave generation sources, communication devices, etc.) on subjects (subjects) and checking communication status
(2) Interference fringe generation and initial position setting of shape reading devices (plural units)
(3) Determination of interference fringe type, color, formation interval, etc.
(4) Shoot the subject with automatic tracking according to the movement of the center of gravity position detection marker.
(5) Based on the interference fringe image, three-dimensional shape data (polygon data) is generated by stereoscopic image processing, and an (outer) shape model of the subject is created.
(6) Equal accuracy and color correction for shape models
(7) A process of attaching a two-dimensional image (surface image) to the shape model.
[0109]
FIG. 23 schematically shows image data G acquired for a subject in a standing posture, and “HEIGHT_tgt” indicates the height of the subject in the diagram shown on the left side. Further, the figure on the right side is obtained by reducing (or expanding) the image data of the subject to the above-mentioned reference height, that is, multiplying the length of all components of the image data G by the ratio of “HEIGHT_ref / HEIGHT_tgt”. Shows an image G ′ obtained by. Note that this ratio “HEIGHT_ref / HEIGHT_tgt” is required in a later step, and therefore needs to be stored in a memory or the like.
[0110]
24 and 25, slice processing was performed at a predetermined interval “ds” along the height direction based on the three-dimensional data (outer shape data and surface state data) acquired for the subject in the standing posture. An example of fault plane sampling is shown.
[0111]
FIG. 24 schematically shows a front view of the subject, and the starting point of the slicing process is the top of the head.
[0112]
As shown in FIG. 25, the area segment ΔS_i (i = 1, 2,...) At each tomographic plane having an identification number i (i = 1, 2,...) Assigned from the side closer to the top to the tip of the foot. For..., Only the shape and cross-sectional area are significant, and there is no data indicating the internal structure related to the area segment ΔS_i. This is because the three-dimensional data of the subject is required and sufficient to be composed of shape data relating to the subject's external shape and data relating to the outer surface state (image data, etc.), and the internal structure of the body is unnecessary It is because it is said.
[0113]
Therefore, the information extracted from the three-dimensional data of the subject is what shape each area segment ΔS_i has and the size of the cross-sectional area (see FIG. 22). In this example, the area segment is obtained by slicing, but instead, a volume segment obtained by multiplying the area segment by the interval ds in the slice direction may be used (in this case, the cross-sectional area Use the volume at the cross section instead.)
[0114]
In addition, the slice interval ds may be set evenly, but a specific portion (abdomen, chest, etc.) characteristically showing the shape of the body is reduced in the slice interval, and the portion is more detailed. Sampling may be performed.
[0115]
Data related to the shape of the subject's body by acquiring the three-dimensional data of the subject by the above method and obtaining data on the shape and cross-sectional area or volume between the tomographic planes from the data. Can be acquired efficiently, and the data input operation can be easily performed.
[0116]
In the next step (3), first, a process for specifying the body shape of the subject is performed.
[0117]
For example, data indicating the shape of each area segment ΔS_i obtained in the step (2) (for example, data indicating the deformation rate of the shape such as the eccentricity when the cross-sectional shape is approximated by an ellipse) is represented by “t_i” (i = 1, 2,...), And the area of the area element ΔS_i is “s_i” (i = 1, 2,...), As shown in FIG. The point PT_i (s_i, t_i) is plotted in polar coordinates (r, θ). That is, in the XY plane, the radius of the circle centered on the origin O indicates the cross-sectional area, and the θ direction indicates the shape of the area segment.
[0118]
27 shows an XY plan view when FIG. 26 is viewed from the Z-axis direction, and points PT_i (s_i, t_i) (i = 1, 2,...) Correspond to each area segment ΔS_i. is doing.
[0119]
After all the positions of the points PT_i are plotted on the XY plane, the center of gravity (referred to as point G) of a polygon (including a convex corner and a concave corner) having each point PT_i as a vertex is obtained. For example, as shown in the figure, a point PT_i, a point PT_ (i + 1), a point PT_ (i + 2) (where i = 1, 2,..., N−2, and n is a natural number) are vertices. The center of gravity of the triangle Gi is the above-described center of gravity G, where “Gi” is the center of gravity of the triangle to be formed. That is, when the position vector of the point Gi with respect to the origin O is a vector “V_Gi” and the position vector of the point G is a vector “V_G”, the vector expression “Σ (V_Gi−V_G) = 0” (where “ "Σ" indicates the total sum for i.) The position of the center of gravity is determined by calculating the coordinates of the point G that satisfies.
[0120]
In FIG. 27, each point PT_i is located in the same quadrant on the XY plane. However, in some cases, the point PT_i is different from the point PT_i as indicated by the points PT ′ and PT ″ in FIG. In the quadrant or in an isolated area away from the area where a large number of points are located together (for example, a fan-shaped area where most points are surrounded by the as axis and at axis) , But some points belong to a fan-shaped area surrounded by the as axis and pi axis.) These points are ignored or treated as exceptions. It is preferable.
[0121]
In addition, it is not always necessary to use all of the area segments (or volume segments) obtained as a result of the above-described sampling, and the sampling results relating to a specific portion (abdomen, chest, etc.) characteristic of the body shape Of course, it is also possible to speed up the processing by selecting only.
[0122]
As described above, after setting polar coordinates (r, θ) on the coordinate plane (XY plane) indicating data on the body shape, the polar angle θ is defined from the shape data (t_i) on the tomographic plane of the subject. The point (PT_i) where the polar radius (r) is defined from the cross-sectional area data (s_i) or the volume data between the tomographic planes (volume data of the volume segment) is defined as the coordinate plane (X− It is possible to specify the body shape of the subject from the position of the center of gravity G of the polygon that is arranged on the Y plane) and connects the points with line segments, and there are four rules for the calculation required for the calculation. Since a calculation amount of about the calculation is sufficient, even if the number of area pieces or volume pieces increases, the computational burden does not increase significantly.
[0123]
When the coordinates of the center of gravity G (which is expressed as “(rg, θg)” in polar coordinate display) are determined, the skeleton data relating to the subject from the functional expression Z = Fn (X, Y) in the body shape coordinate space described above. The conversion ratio (the ratio used for creating the model related to the target person) can be calculated. That is, as described above, the functional equation Z = Fn (X, Y) is obtained from the interpolation calculation based on the skeleton data of each reference body type, and therefore, using the conversion equation from the polar coordinate system to the two-dimensional orthogonal coordinate system, “Xg = By calculating “rg · cos (θg)” and “Yg = rg · sin (θg)” and substituting them into the function formula, the value of Fn (Xg, Yg) can be obtained.
[0124]
FIG. 28 conceptually shows how a function value is obtained from the center of gravity G (Xg, Yg). For Fn (X, Y), n is selected as “_hr_xx” and n is set to “_mr_xx”. "Is shown together with the case of selecting. That is, the height (Z_hr) of the point Q_hr indicates the value (_hr_xx) of the Fn (Xg, Yg) related to the length ratio, and the height (Z_mr) of the point Q_mr is related to the weight ratio Fn (Xg, Yg). ) Value (_mr_xx).
[0125]
Thus, when the conversion rate of the subject is obtained for the length and weight of the bone, the actual length and weight can be calculated based on the height and weight of the subject, and this processing is performed in the next step (4). Is called.
[0126]
For example, the functional expression Fn (X, Y) related to the ratio of length is a value obtained when the height of each reference body model described above is aligned with the reference height and the height of the subject is converted to the reference height. Therefore, the above-mentioned ratio “HEIGHT_ref / HEIGHT_tgt” is required to calculate the bone length with respect to the height of the subject. That is, the length of the portion indicated by xx is determined by multiplying “_hr_xx” by HEIGHT_tgt. It should be noted that other quantities having a dimension of length, such as bone thickness, can be obtained in the same manner as in the derivation process of _hr_xx.
[0127]
As for the weight ratio, the weight of the part indicated by xx is determined by multiplying the subject's weight (body weight, etc.) by “_mr_xx”, and the length of the cross section, volume, etc. is raised to the nth power (where n is 2 or more). An amount having a dimension of a natural number.) Can be obtained in the same manner as in the weight deriving process.
[0128]
In this way, a numerical model relating to the skeleton (structure) of the subject can be created by determining the length and weight of all bones related to the subject. For example, when a model based on polygon data is adopted as the representation form of the numerical model. Can prepare polygon data of the whole bone in advance and use a deformation process such as morphing to obtain a three-dimensional model of the skeletal structure.
[0129]
In the above description, the three-dimensional data (outer shape data) of the subject is acquired and the data related to the body shape of the subject is obtained, but the two-dimensional image data of the subject is obtained without using the three-dimensional data. In the case of creating a simple model by using only the length ratio (the aspect ratio, etc.) of the human body characteristically showing the body shape from the image data, the cross-sectional area of the above-described area segment may be substituted. Of course. For example, the image data of the subject in a standing posture is typically a warrior-type inverted triangle (inverted trapezoidal) shape, an obese type is almost trapezoidal, and a lean type is almost rectangular. Can obtain information on body shape.
[0130]
On the other hand, in order to improve the accuracy of the model, the data is corrected based on the length of each part of the human body obtained from the three-dimensional data of the subject, or the total bone weight exceeds the body weight. Therefore, it is necessary to reduce the error between the structural model and the actual target human body as much as possible so as to prevent inconsistency.
[0131]
Next, creation of a numerical model obtained by adding the above (ii) muscle model to the skeleton model will be described.
[0132]
In this case, in addition to the shape, length, and weight of the muscle, various quantities related to the exercise performance of the muscle (for example, muscle contraction rate, work rate, muscle contraction response speed, etc.) should be included in the numerical model. It is preferable to quantify the performance of muscles.
[0133]
That is, the length and weight of the muscle can be obtained by following the same procedure as “_hr_xx” and “_mr_xx” described above. That is, “_hr_xx” indicates the ratio of the length of each muscle indicated by xx, and “_mr_xx” indicates the ratio of the weight of each muscle indicated by xx.
[0134]
On the other hand, the muscle contraction rate and work rate are determined depending on what is affected by these values.
[0135]
For example, the muscle contraction rate is a ratio indicating how much the muscle contracts with respect to the natural length of the muscle, the reference length (natural length) of the muscle is represented by a variable “L”, and the muscle length at the time of the muscle contraction is expressed. When expressed by the variable “LL”, it can be expressed by a function expression f (L, LL) (in the simple model, “f (L, LL) = LL / L”). Similar to the function Fn (X, Y) described above, it can be calculated from the data of each reference body model. Therefore, “Lg = Fn (Xg, Yg), LLg = Fm (Xg, Yg)” (where Fn represents a function when the Z axis is taken as a variable L, and Fm represents the Z axis as a variable LL. The muscle contraction rate of the subject can be calculated from f (Lg, LLg) obtained by substituting the value calculated from the above formula into the above function formula.
[0136]
Similarly, the work rate (work per unit time) can be obtained as a function of the weight of the action point and the action time, and the reaction rate of muscle contraction is the distance from the terminal nerve to the muscle and the start of muscle contraction. As a function of time.
[0137]
In addition, about the creation process of the model containing a skeleton and muscles, the skeleton and muscles including the shape, length, weight and muscle contraction rate of the muscle in addition to the shape, length and weight of the bone in the step (1) described above. Are prepared, and in the step (3), interpolation calculation is performed based on the skeleton and muscle data related to each reference body type, and the conversion ratio of the skeleton and muscle data related to the object is calculated. In step (4), the length, weight, and muscle contraction of each bone and muscle of the target based on the conversion ratio obtained in step (3) and the length and weight of the target input in step (2). The rate should be determined.
[0138]
According to the model created in this way, for example, it is possible to show a simulated exercise of muscles when a bone, a joint, or the like is moved.
[0139]
This is the end of the description of the skeleton and muscle numerical model. Since the above method has universality that can be adopted regardless of the numerical data, this method is described in (iii) to (iii) above. It is easy to apply to the model of (v).
[0140]
Next, a database constituting the human body structure model will be described. Here, the “human body structure model” is a numerical value constructed on a computer as a structural model including the skeleton, muscles, nerves, skin, fat, etc. from the basic data such as the physique and weight of the human body. Mean model.
[0141]
This human body structure model is constructed from a large number of DBs. When the DBs are roughly classified, the following DBs are conceptually listed.
[0142]
(A) DB related to bone
(B) DB for joints and ligaments
(C) DB related to muscle
(D) DB for nerves (motor nerves, reflexes, etc.)
(E) DB related to internal organs and fat
(F) DB for skin
(G) DB related to physical exercise
First, (A) includes items such as bone weight, weight distribution, shape, and fracture coefficient, and is a DB related to basic information of the skeleton.
[0143]
Also, (B) includes information on the degree of freedom of joints and the breaking coefficient, the connection relationship between the even number and the even number, the connection relationship between the even number and the muscle, the relationship between the contraction rate of the muscle and the angle of the joint, and the like.
[0144]
(C) includes information on the free length of the muscle, the length during contraction (or expansion), the contraction rate, the breaking coefficient, the weight, the reaction rate, the work rate, and the like.
[0145]
(D) includes, for example, information regarding the relationship between motor nerve signals and muscle contraction in the brainstem, central nerve, terminal nerve, and muscle contraction nerve regarding motor nerves, or tactile nerve, terminal nerve, and central nerve regarding reflex nerves. , Information on the relationship between pressure, heat, pain sensory nerve signals and muscle contraction in the brainstem.
[0146]
(E) includes information on the weight and weight distribution of viscera and fat, and (F) includes information on skin aging and wrinkle leaning.
[0147]
(G) is a DB for representing only the dynamic motion state of the human body, which includes data obtained by extracting information on the whole body motion and partial motion of the human body as a wire frame model, and the position of the center of gravity accompanying the motion. Contains information about changes.
[0148]
Each of the above DBs is not always generated as one DB (for example, a certain DB is configured as a set of a plurality of DBs).
[0149]
The summary of the processing related to the creation of the human body structure model can be summarized as an itemized list as follows.
[0150]
Step SS1: Input basic data about the body
Step SS2: Data processing based on DB
Step SS3: Acquisition of three-dimensional data about the human body and generation of a wire frame model relating to movement
Step SS4: Acquisition of three-dimensional data about the human body and acquisition of center-of-gravity position data accompanying exercise
Step SS5: Acquisition of three-dimensional data about the human body and acquisition of balance nerve signals and body balance by the semicircular canal nerve
Step SS6: Acquisition of 3D data about the human body and acquisition of relationship between motor nerve signals and muscle contraction by electromyogram etc.
Step SS7: Acquisition of 3D data about the human body and acquisition of the relationship between muscle contraction and nerve signals of pressure, heat and pain sensation by reflex nerve test
Step SS8: Acquisition of electromyogram information on stretch reflexes.
[0151]
Of the above steps, steps SS1 to SS4 related to the present invention will be described below.
[0152]
The SS1 is a step of inputting data such as the physique, gender, and age of an individual subject to a human body structure model. In the subsequent step SS2, the data is processed based on various DBs and related to the body structure of the subject. DB is generated.
[0153]
Here, various DBs are, for example, as shown in the following (a) to (f).
[0154]
(A) Weight DB
It is a DB relating to the weight ratio (ratio to body weight) of bones, muscles, fats, heads, organs, and the like.
[0155]
(B) Weight distribution DB
This is a DB related to weight distribution (position, center of gravity, etc. in the human body) of bones, muscles, fats, heads, organs, and the like.
[0156]
(C) Failure factor DB
It is a DB related to a fracture coefficient (such as a fracture coefficient) of bones, ligaments, muscles and the like.
[0157]
(D) Joint degree of freedom DB
It is a DB related to the degree of freedom of each joint, and is used for setting the movable range of the joint.
[0158]
(E) Connection relation DB
It is a DB that prescribes connection relations between an even number and an even number, between an even number and a muscle, between an even number and a ligament.
[0159]
(F) Muscle movement DB
This is a DB relating to the joint drive angle and the drive ratio of each working muscle, the free length of muscle, the contraction rate, the work rate, the reaction rate of muscle contraction, and the like.
[0160]
(G) Nerve placement DB
This is a DB related to nerve arrangement and length in the human body.
[0161]
Also, a DB related to body structure (hereinafter referred to as “body structure DB”) is defined as a DB generated by the link (data combination or association) of the DBs (a) to (f).
[0162]
Table 1 below shows an example of the relationship between the DBs (a) to (f) and the DBs (A), (B), (C), and (E). It is.
[0163]
[Table 1]

[0164]
In Table 1, “◯” means that each DB shown in the horizontal column includes the DB shown in the vertical column, and “−” means that there is no such inclusion relationship. For example, the DB related to the bone in (A) includes a DB related to the weight and weight distribution of the bone and the fracture coefficient of the bone. In addition, with regard to DB (B), the weight of the ligament and the weight distribution are ignored because the weight of the joint is mainly composed of the weight of the bone part. Of course, it can be realized.
[0165]
The DB related to the shape is generated from the body shape, sex data, and the like using the above-described method based on the shape model of the skeleton, fat, and the like of the reference human body (the above-mentioned reference body model etc.).
[0166]
FIG. 29 and FIG. 30 are flowcharts showing the main part of a specific example of the above steps SS1 and SS2.
[0167]
First, for the following data items defined in advance in step S1 of FIG. 29, numerical values of the subject are input (manually or automatically input), or in the case of a selection value, any of them is designated.
[0168]
・ Height (unit: mm)
・ Weight (Unit: Kg)
・ Body shape (types such as lean type, warrior type, obese type, etc.)
·Gender Male Female)
・ Age (unit: years old).
[0169]
For example, the example shown in the following table 2 about the data structure of an input value is given.
[0170]
[Table 2]

[0171]
In other words, in this case, a value indicating the type (such as the value of the coordinate axis θ in the body shape coordinate space described above) such as those requiring an area for storing continuous numerical values such as height and weight, or the body shape, etc. Corresponding) and a numerical value indicating the degree (corresponding to the value of the coordinate axis r in the body shape coordinate space) or simply expressed by 1-bit data of “0” or “1” like gender What can be mentioned.
[0172]
When generating a body structure model, the model generated based on a relatively small number of input parameters as described above (hereinafter referred to as “normal mode”) and the model generated in the normal mode are modified. Mode (hereinafter referred to as “special mode”), the process in steps S3 to S8 will be described first assuming that the normal mode is selected in the mode determination process in step S2.
[0173]
In step S3, a weight setting process is performed with reference to the weight DB of (a). In other words, the weight ratio for bone, muscle, fat, head, organ, etc. is set based on the body shape data of the subject, and the weight of muscle and fat is further divided into the chest, abdomen, upper limbs, and lower limbs. Set. Moreover, since the way of attaching muscle fat differs depending on gender, the weight ratio is set in consideration of the difference. Of course, it is necessary to assign the weight so that the difference between the total weight after the setting and the input value of the body weight is substantially zero.
[0174]
In the next step S4, with reference to the weight distribution DB in (b) above, the skeleton and fat shape are set based on the height and body shape data of the subject. Then, the bone and head weight distribution is set by the bone weight distribution (center of gravity, specific gravity, etc.), and the muscle and fat weight distribution is set by the flesh weight distribution. Here, the “weight distribution of meat quality” will be described separately for the case of the upper limb, the lower limb, and the case of the chest. In the former case, a virtual circle centered on the weight point of the bone weight distribution. Set multiple orbits (this is the shape of a human body in a standing position when viewed from the front, and more precisely it has a spherical shape.) To distribute. In the latter case, a plurality of virtual trajectories are set at a predetermined distance from the weight point of the rib cage in the bone weight distribution to the outside, and the weights are evenly spaced with respect to each trajectory. To distribute. If the gender input data is female, an additional setting is made for the weight distribution of chest fat.
[0175]
The visceral weight distribution is set at equal intervals on the visceral weight line connecting the head visceral part, the first intercervical joint and the pubic bone, and the weight distribution changes according to the movement of the vertebrae. As for the weight distribution of abdominal fat, a virtual elliptical trajectory centered on weight points set at equal intervals along the visceral weight line (this is the shape when a human body in a standing posture is viewed from the front. Yes, it has an ellipsoidal shape), and a plurality of weights are evenly distributed at equal intervals for each track. In addition to this, there is a method in which bone, muscle, internal, and fat regions are provided in a virtual elliptical orbit to create a DB for each region, and an equally distributed weight is set for each region.
[0176]
Since the visceral and fat distributions are clarified as described above, it is possible to set an even entry prohibition range (a range for prohibiting the even couple from entering the body through the internal organs).
[0177]
In step S5, the fracture coefficient data for bones, ligaments, muscles, etc. are read from the fracture coefficient DB of (c) above. This is necessary in order to avoid a posture in which an excessive force that causes damage to bones, ligaments, and the like, that is, a human body state that is impossible in a healthy body, is avoided in the human body structure model.
[0178]
In step S6, with reference to the joint degree of freedom DB in (d) and the destruction coefficient DB in (c), the joint drive range and degree of freedom (including the degree of freedom at the time of destruction) from the data of body shape and sex. Set. This is to exclude movements that are not allowed as a healthy body or impossible movements from the joint structure.
[0179]
In step S7, after referring to the above (e) connection relation DB and reading data on various connection relations, in the next step S8, settings for muscle movement are performed.
[0180]
For muscle movement, referring to the muscle movement DB in (f) above, the joint drive angle and the weight of the kinematic pair driven based on the input data (height, weight, figure, sex, age, etc.) Set the contraction rate of the working muscle.
[0181]
Through the above steps S3 to S8, the body structure model in the normal mode and the body structure DB in which this is made into a database are created. Note that the nerve arrangement DB (g) is created by a single processing process from the nerve arrangement of the reference human body (reference body model etc.) in accordance with the body structure DB.
[0182]
In step S9 of FIG. 30, it is determined whether or not to proceed to the special mode. If the special mode is selected, the process proceeds to step S10, and if not selected, the process proceeds to step S17.
[0183]
When the special mode is selected in step S2 or step S9, the process proceeds to step S10, and it is determined whether or not the ratio conversion mode is selected. The “ratio conversion mode” is a mode in which the size of the arm, torso, leg, etc. is input to the data generated in the normal mode, the data is compared, and the length and weight are reset. . If the ratio conversion mode is selected in the same step, the process proceeds to step S11, and if not selected, the process proceeds to step S17.
[0184]
In step S11, the ratio conversion mode is further divided into two modes, that is, a “length setting mode” and a “ratio setting mode”, and one of the two is selected. If the “length setting mode” is selected, the process proceeds to step S12, and the lengths (unit: mm) of the arms, torso, legs, etc. are input. If the “ratio setting mode” is selected, the process proceeds to step S13, and ratios with respect to reference lengths (eg, height, etc.) of the arms, torso, legs, etc. are input. Note that the input data in steps S12 and S13 is used, for example, for correction of the free length of the muscle in the connection relation DB (e).
[0185]
In subsequent step S14, the weight mode is divided into two modes, that is, the “matching mode” and the “adjustment / adjustment mode”, and one of the two is selected. Then, if the “matching mode” is selected, the process proceeds to step S15, and after processing for matching the weight obtained from the generated data with the input data of the body weight, the process returns to step S2 in FIG. When the “adjustment / adjustment adjustment mode” is set, the process proceeds to step S16 to perform a process of giving an increase / decrease amount for the weight obtained from the generated data, and then returns to step S2. Through these steps S15 and S16, the weight setting and the weight distribution correction in the above steps S3 and S4 can be further performed.
[0186]
In step S17, it is determined whether or not the “joint weight mode” is selected. If so, the process proceeds to step S18. If not selected, the process proceeds to step S19. The “joint weight mode” is used when, for example, the weight of a joint driving motor or the like is registered in a biped robot that imitates a human body instead of a human body. This is a mode for setting the kinematic weight distribution so that the overall balance is maintained by adding and subtracting the weight values. Note that after the processing in step S18, the process proceeds to step S19.
[0187]
In step S19, it is determined whether or not the “muscle weight mode” is selected. If so, the process proceeds to step S20, where a desired weight value is added or subtracted for each muscle or fat weight associated with each pair. By doing so, it is set so that the weight balance of muscles and the like is maintained.
[0188]
If the “muscle weight mode” is not selected, the generation of the body structure model is terminated. However, only a relatively small amount of data relating to the physique, sex, etc. is necessary in the steps so far.
[0189]
Next, steps SS3 and SS4 described above will be described with reference to the flowchart of the main part shown in FIG. In this step, the following DB is created for the above-mentioned (G) DB relating to physical exercise, and the contents thereof will become clear from the description of the processing described later.
[0190]
(G1) Exercise wire frame DB
(G2) Motion center-of-gravity position DB.
[0191]
First, in step S1 of FIG. 31, the subject who is provided with the center-of-gravity position detection marker is maintained in the standing posture, and then in the next step S2, the three-dimensional data acquisition method is used. Get the data. Here, “three-dimensional data” is data that is distinguished from a two-dimensional image pasted on a plane, and means a curved surface forming a three-dimensional shape and image data pasted on the curved surface. .
[0192]
Any method can be used as a method for obtaining three-dimensional data as long as it can obtain data including the shape and image of the subject. For example, a method using the above-described interference fringes of light is exemplified. be able to.
[0193]
In the next step S3, the model obtained in the previous step is sliced along the vertical direction from the head toward the toes (that is, a tomographic cross section is formed), whereby each component of the body. Recognize and even numbers. For example, with respect to the model mdl schematically shown in FIG. 32, the direction indicated by the arrow R is the slice direction, the point Ps indicates the crown, the region “A1” is the region related to head recognition, and the region “A2” is the cervical spine. A region related to recognition, a region “A3” is a region related to shoulder recognition, a region “A4” is a region related to chest and abdomen recognition, and a region “A5” is a region related to leg recognition. Note that the tomogram schematically shown on the right side in FIG. 32 has one region sliced at the lower boundary surface of the region A1, and three regions sliced at the lower boundary surface of the region A3. This indicates that the number of areas obtained by slicing the part in the middle is two.
[0194]
The kinematic pair is recognized by comparing and contrasting the subject's body structure model with a model obtained from three-dimensional data. For example, the axis connecting the center of gravity corresponding to the center-of-gravity position detection marker attached to the subject (subject) and the center point of the head is recognized as the backbone axis of the body structure model, or the positions of the shoulder joints of the upper limbs Can be determined on the basis of the physical characteristics of the human body, such as being recognized as being the most protrusive part with respect to the spine axis.
[0195]
In step S4 of FIG. 31, the subject is caused to perform knee flexion and extension, and processing relating to acquisition of three-dimensional data and knee joint recognition is performed in the same manner as described in steps S2 and S3. In other words, it is because the knee joint cannot be accurately identified only by the three-dimensional data in the standing posture.
[0196]
In step S5, the subject moves his / her arm so that his forearm is parallel to the ground, and the three-dimensional data acquisition and elbow joint recognition are performed in the same manner as described in steps S2 and S3. Process.
[0197]
In the next step S6, the basic data obtained in the above steps S2 to S5 is corrected (error correction, etc.), and then the process proceeds to the next step S7, where data is determined, that is, human body shape data relating to the subject is determined. To do.
[0198]
Then, the process proceeds to step S8, and a fitting process between the shapes of the subject and the human body shape data obtained in the above step is performed so as not to cause a significant difference. Minimize differences between models.
[0199]
In step S9, the subject is given a predetermined exercise, and three-dimensional data regarding the exercise state is acquired using the same method as described in step S2.
[0200]
In step S10, the body frame is recognized based on the positions of the head and the center-of-gravity position detection marker in the three-dimensional data, and a wire frame is used as a frame model that represents information related to the subject's physical movements (such as the state of the even number). Generate a model. Then, the wire frame model is created as a database to create the motion wire frame DB.
[0201]
In step S11, based on the wire frame model and the human body shape obtained in the previous step, only the center of gravity position, which is the center of motion, and data relating to the change are extracted to create the motion center of gravity position DB.
[0202]
Thus, numerical models related to the structure and movement of the body can be obtained by the processing that has been performed so far, and by adding polygon data (data by polygon approximation or polyhedron approximation) about the shape of muscles and fats to these models A polygon model related to the human body and a database of the polygon model (hereinafter referred to as “human body polygon DB”) can be obtained.
[0203]
FIG. 33 shows an example of the dependency of the human body polygon DB with the following DB in addition to the above-described body structure DB, exercise wire frame DB, exercise center of gravity position DB, and an arrow “→” in FIG. , “X → Y” means that the database “Y” is generated based on the database “X”, and both arrows mean links.
[0204]
(H) Muscle polygon DB
(I) Fat polygon DB
(J) Special fat polygon DB
(K) Skin aging polygon DB.
[0205]
The DBs (h) to (j) will be briefly described. First, (h) the muscle polygon DB is a collection of polygon data of muscle shape corresponding to muscle contraction, and (i) the fat polygon DB is a muscle polygon database. It is a collection of fat-shaped polygon data corresponding to the contraction and the movement of the center of gravity. (J) The special fat polygon DB is a collection of polygon data on the shape of fat (visceral fat, chest fat, etc.) that is not directly related to muscle contraction but is mainly involved in the movement of the center of gravity.
[0206]
(K) The skin aging polygon DB is a polygon data collection (or two-dimensional image data) regarding the amount of wrinkles required for artificially aging the skin and how to approach the skin. ) It constitutes a DB related to skin.
[0207]
The “muscle contraction polygon DB” shown in FIG. 33 is a database generated from the muscle polygon DB, the body structure DB, the exercise wire frame DB, and the exercise center of gravity position DB, and expresses the contraction state of the muscle accompanying the movement of the human body. It is necessary to do this and is generally generated according to the following procedure.
[0208]
(1) Generation of body structure DB
(2) Generation of motion wire frame DB and motion center of gravity position DB
(3) Link between muscle polygon DB and (1) and (2) DB.
[0209]
The “fat contraction polygon DB” is a database generated from the fat polygon DB and the special fat polygon DB with reference to the muscle polygon DB, the body structure DB, the exercise wire frame DB, and the exercise center of gravity position DB. It is required to express changes in the position and thickness of fat accompanying the movement of the center of gravity.
[0210]
FIG. 34 is a flowchart showing a main part of a processing example for generating a human body polygon DB, and shows processing related to generation of a muscle contraction polygon DB and processing related to a fat contraction polygon DB in parallel.
[0211]
First, regarding the processing related to the generation of the muscle contraction polygon DB, after preparing the muscle polygon DB, the body structure DB, the exercise wire frame DB, and the exercise center of gravity position DB in step S1, in step S2, the human body movement and the muscles of each muscle are prepared. Obtain a relationship for the change in shrinkage.
[0212]
In step S3, a polygon data collection showing muscle shapes corresponding to the contraction of all muscles, that is, a muscle contraction polygon DB is created. At that time, the data related to the human body shape and the data of the muscle polygon DB are referred to in the body structure DB.
[0213]
In the next step S4, after adding the polygon data of the muscle part to the exercise wire frame DB, the process proceeds to step S5.
[0214]
On the other hand, regarding the processing related to the polygon DB of fat and special fat, after preparing both DBs in step ST1, the relationship between the human body motion and the change in contraction rate of each fat is obtained in the next step ST2.
[0215]
Then, in step ST3, a polygon data collection showing fat shapes corresponding to muscle contraction and center of gravity movement for all fats, that is, a fat contraction polygon DB is created. At that time, it is necessary to refer to the data related to the human body shape and the data of the muscle polygon DB in the body structure DB.
[0216]
In the next step ST4, after the fat part polygon data is added to the exercise wire frame DB, the process proceeds to step S5.
[0217]
In step S5, for an exercise wire frame model that takes into account muscles and fats, muscle muscles and fat polygonal data are processed for each frame of the image data to express muscle movement, fat shaking, and the like. This processing is performed as data processing for dots and patches on the surface of the polygon, and includes, for example, connection processing and bezier processing for dots, or processing such as processing after compression of a certain surface.
[0218]
In the next step S6, the skin aging polygon DB and the motion wire frame DB are linked (indicated by arrows in both directions in FIG. 33). Thereby, for example, the skin wrinkles around a certain joint can be changed according to the joint angle and the distance from the joint.
[0219]
In this way, the human body polygon DB is created. For electrical stimulation to muscles, for example, the following procedures (a) to (c) are followed.
[0220]
(B) Identify the joint to be driven
(B) Select the muscles necessary to operate the joints in (a) (for example, extensors and flexors as the paired muscles) or the muscles to be stimulated.
(C) In the muscles selected in (b), the surface layer near the skin surface is stimulated by a low frequency (or stimulation by phantom sensation described later) through the effector, and is effective for the inner layer muscle. Giving stimulus by interference wave through the vessel.
[0221]
In other words, by properly using the mode of electrical stimulation (interference wave electrical stimulation or low-frequency electrical stimulation) according to the position of the muscle to which the stimulus is applied, the device efficiently gives a stimulus to each muscle and produces a sense of incongruity. It can be presented to the user.
[0222]
In the above procedure (A), the joint to be driven is specified. However, depending on the effector, the target joint related to the effector may already be specified. Can do.
[0223]
In (b), when the extensor and flexor muscles are selected as the muscles necessary for the movement of the joint, the extensor and flexor muscles are not contracted to the same extent. (For example, the ratio of flexion to extensor contraction is about 20: 1.) On the contrary, the extensor is contracted more when the joint is extended. In the selection of muscles, a database that associates joints with the expansion and contraction of the muscles that drive the joints is built in advance, and when the target joint is specified in the process (a) above, It is preferable to select a muscle necessary for driving the joint based on the database. The “database that relates the joint and the expansion and contraction of the muscle that drives the joint” can be generated by, for example, associating the DB related to (B) the joint or the ligament and (C) the DB related to the muscle. Among the DBs shown in (a) to (f), the data items for the joints and the kinematic pair and the muscles supporting the joint (for example, the name and position of the joint, the name and shape data of the muscle required for driving the joint, the kinematic pair The joint name of the starting point that supports When this database is used, for example, when the elbow joint is targeted, information on the muscles involved can be obtained according to the movement.
[0224]
For example, the drive of the elbow joint will be described as follows.
[0225]
Among the muscles involved in the movement of the elbow joint and forearm in the human body, the upper arm muscles include the biceps, humerus, triceps, and elbows. , Supination muscle, circular pronation muscle, square pronation muscle, and the like.
[0226]
Among these, when looking at the muscles of the upper arm, the biceps, which are the two joint muscles, have a bending action on the elbow joint and a supination action on the forearm, and in particular, bending has a strong function in the flexor muscle group. It is known. The triceps surae is a biarticular muscle with a long head starting from the scapula, and the forearm extension at the elbow joint is performed mainly by the muscles belonging to the medial head, and has a strong extension force. Are known.
[0227]
When mounting the above-described multi-layer effector on the muscle, efficiency can be improved by arranging the electrode group only in a specific region rather than uniformly arranging the electrode group over the entire sheet surface. preferable. This is because the influence of electrical stimulation varies depending on the location of the muscle. The “specific region” refers to a region on the sheet-like substrate corresponding to a muscle start point or muscle abdomen, or a region in the vicinity thereof or a range extending from the muscle start point to the end point. For example, in the case of the biceps brachii muscle, in this case, the electrode placed on the muscle abdomen is the negative electrode, the electrode placed on the muscle start point is the positive electrode, and a pair of electrodes are used. By applying low-frequency electrical stimulation, it is possible to provide stimulation to the surface muscles whose muscles can be visually confirmed from the skin surface.
[0228]
Moreover, it is preferable not to arrange | position the electrode group which comprises an electrode part in the area | region of the skin surface which is caught by operation | movement of a joint. The reason is that the movement of the electrode position accompanying the deformation of the sheet-like base material is large, and when the number of the resistance patterns, the number of pressure and temperature detection elements is large, the processing of the acquired data is burdened. Because. Therefore, in order to avoid such an inconvenience, as shown in the following (i) to (iv), a region where the pressure changes greatly during the joint operation (bending, stretching, etc.) is examined in advance. Is preferred.
[0229]
(I) Attach effector to subject
(Ii) Ask the subject to perform certain movements of joint flexion and extension
(Iii) A location where the resistance value change or surface pressure information of the resistance pattern during the operation of (ii) is acquired and a predetermined threshold is set for the resistance value or pressure value to indicate a position or range beyond this Information is converted into data and stored in storage means such as a ROM (read-only memory) or an auxiliary storage device (or external storage device).
(Iv) With reference to the location information stored in (iii), the arrangement of the electrodes used and the area where pressure detection is not performed (or the detection value should be ignored) are determined.
[0230]
In (ii), it is preferable to use a method in which a visual display device such as a head-mounted display (HMD) is mounted on the subject and the operation to be imitated is displayed on the device. In other words, a video that shows image information obtained by camera shooting as a moving image and a virtual video that shows an operation to be imitated are displayed on a visual display device, and a video showing the actual operation of the subject is virtually displayed to the target person. If you follow a certain movement by following the movement so that it overlaps the video, it is more efficient than imitating the verbal explanation or the action of the instructor.
[0231]
Further, when the effector is worn and used in a shape that can be used, a material having breathability and stretchability is used in a place where the electrode group constituting the electrode portion is not arranged in the sheet-like base material. It is preferable. That is, it is desirable to use a material that is highly breathable and has excellent stretchability in order to cope with heat generation and sweating associated with the movement of the subject (that is, to prevent stuffiness) and to ensure ease of movement. For example, Darlington stretch fabric and a fabric made by laminating a thermoplastic film with water vapor passing through it, LeBand's “Darlex” (trademark) and a cotton material with excellent hygroscopicity and water supply properties Mizuno's “Ice Touch” (trademark), etc., is a functional material made by combining high-priced Eval fibers (composite fiber using polyester for the core and ethylene / vinyl alcohol for the sheath: Kuraray's trademark). However, the present invention is not limited thereto, and materials developed in the field of sports clothing and the like can be used.
[0232]
Then, instead of covering the joint and the entire portion driven by the effector with the effector having the electrode part only in the part corresponding to the kinematic pair about the joint (i.e., While creating an electrode part for every pair, wear the effector of the shape according to the shape characteristic of each pair.) By giving electrical stimulation, it becomes light and easy to wear. For example, in the case of the elbow joint of the forearm and the wrist joint, for example, it is only necessary to attach each effector in the vicinity of these joints. Note that the shape recognition of the even number will be described later.
[0233]
Thus, the basic procedure of the control method according to the present invention is summarized as an itemized list as follows.
[0234]
(1) Build in advance a database showing the arrangement of human muscles related to the subject
(2) Wear the effector on the subject and set the electrode group so that it contacts the skin surface.
(3) Of the electrodes constituting the electrode group of (2), the frequency component related to the electrical stimulation signal to the electrode set as the anode side electrode and the frequency of the signal detected from the electrode set as the cathode side electrode By calculating the difference data between the components, the frequency component of the signal generated from the muscle by the electrical stimulation is analyzed, and the difference data is used as the data of the frequency component related to the action set state signal of the multi-muscle fiber by ATP. Save
(4) Recognize the shape by acquiring information indicating the shape and shape change of the skin surface
(5) Select an electrode to which an electrical stimulation signal should be supplied from the electrode group of the effector based on the information on the muscle arrangement of the subject obtained from the database in (1) and the shape recognition result obtained in (4)
(6) By comparing the myoelectric potential information detected from the electrode selected in (5) with the data stored in (3), the operation is predicted before the operation of the muscle, and then the electrode An electrical signal for low-frequency electrical stimulation or interference electrical stimulation is supplied to.
[0235]
The outline explanation is finished above, and the basic items shown below are explained in detail.
[0236]
(I) Shape recognition method using pressure detection
(II) Construction of database using information on bone cross section and muscle and fat cross section
(III) Low frequency electrical stimulation method and interference wave electrical stimulation method
(IV) Phantom sensation
(V) Motion prediction method using electromyogram
(VI) Prevention of low-temperature burns and adjustment of stimulus output
First, as described above, the shape recognition unit 1c in FIG. 1 has been described with respect to the recognition method using pressure detection. However, in this method, a numerical relationship between the surface pressure P and the shape S is acquired in advance. The initial learning process for this purpose will be described with reference to FIGS.
[0237]
FIG. 35 shows a pressure detection sheet 12 formed by laminating row and column electrode sheets and a pressure-sensitive layer on a sheet-like base material, and is rounded into a cylindrical shape (or a truncated cone shape), and ends thereof are made of silicone or rubber material. A bonded state using a cloth or the like is shown (the hatched portion in the figure indicates the bonded portion 13). In addition to this, as shown in FIG. 36, the adhesion region 14 may be formed so as to cover the entire side surface of the cylinder (pressure detection sheet 12) (see the hatched portion in the figure). ). In addition, it is preferable to provide warp yarn (for example, strong industrial yarn such as hemp) on the inside or the outer surface of the adhesive part or the sheet base so as to follow the generatrix on the side surface of the cylinder (sheet base in a direction parallel to the cylinder axis) However, when the electrode sheet or pressure sensitive sheet is deformed when the pressure detection sheet is extended, the resistance value, dielectric constant, etc. may change. It is preferable to measure the change rate of the physical quantity in advance and perform data correction in consideration of the change rate.
[0238]
FIG. 37 schematically shows a configuration for selecting each pressure detection element on the pressure detection sheet, and the matrix selection circuit 15 is connected to the row electrode groups 8a, 8a,... And a connection line group “Y” connected to the column electrode groups 10a, 10a,..., And an address bus 18 related to the CPU 16 and the memory 17 (RAM or ROM). And connected to the data bus 19. That is, when the address designation for the pressure detection element group is performed by the address designation from the CPU 16 to the matrix selection circuit 15, the corresponding electrode of the row electrode group and the column electrode group corresponding to the address is selected. The position of the pressure detection element is specified. Then, detection information of the selected pressure detection element is returned and used in the CPU 16 and the memory 17 via the data bus 19.
[0239]
FIG. 38 shows a reference column (a cylinder may be used at the time of initial learning, but it is necessary that the shape does not easily change due to a force received from the sheet when the cylinder is inserted into the pressure detection sheet). In this example, three cylinders α, β, and γ having different radii are prepared, and the radius “r_α” of the column α is the smallest among the three, and the radius “r_β” of the next cylinder β is the smallest. Is larger than the radius of the cylinder α, and the radius “r_γ” of the cylinder γ is the largest diameter.
[0240]
Each of these reference cylinders is inserted into the pressure detection sheet 12 (in the figure, has a truncated cone shape), and the numerical relationship between the pressure distribution and the shape (radius of the reference cylinder) is obtained by calibration. To do. In other words, it is clear that the larger the diameter of the reference cylinder, the greater the pressure when it is inserted into the cylindrical pressure detection sheet (the surface pressure increases), and the pressure detection information at that time (exactly Can obtain the correspondence between the data group obtained by each pressure detection element) and the shape as numerical information.
[0241]
In this example, for simplicity of explanation, the number of reference cylinders is three, so the number of points for determining the free curve is small, but a large number of reference cylinders with different radii are prepared or one The number of data necessary for interpolation is secured by providing a mechanism for continuously changing the radius of the reference cylinder (for example, artificial muscles using pneumatic pressure, electromagnetic mechanism, etc. are attached to the reference cylinder). Of course you can. As for the shape of the reference cylinder, a quadrangular column or a polygonal column may be used in addition to the column.
[0242]
FIG. 39 is a flowchart showing a procedure example of initial learning. First, in step S1, one of the reference cylinders (or reference cylinders, prisms, etc.) is selected, and this is used as the cylindrical pressure detection sheet 12. insert.
[0243]
In the next step S2, each pressure detection element in the pressure detection sheet is scanned by the matrix selection circuit 15 to acquire pressure detection information by these, and store it in a memory (RAM).
[0244]
In this case, in order to cope with the inconvenience that the amount of data increases due to the huge number of pressure detecting elements, the above-mentioned “numerical model of mechanical structure related to human body” is used. By selecting the minimum necessary number of pressure detection points, the amount of data can be greatly reduced. In other words, in order to reduce the amount of data, it may be possible to reduce the number of pressure sensing elements to the minimum necessary in advance, but this will result in individual differences among subjects (physique differences, limb length differences, etc.) This is because the pressure detection point cannot be selected at an appropriate position.
[0245]
In the next step S3, it is determined whether or not the operation for obtaining the pressure distribution data by inserting these into the cylindrical pressure detection sheet is completed for the prepared reference cylinder and the like proceeds to the next step S4 at the end. Otherwise, the process returns to step S1.
[0246]
In step S4, in order to obtain a functional relationship between the shape S and the pressure P for each pressure detection point, an approximation process is performed using a free curve or a quadratic curve. For example, the free curve includes a Lagrange interpolation curve, a Ferguson curve, a spline interpolation curve, a Bezier polynomial curve, a B (Bézier) -spline curve, etc., and it is preferable to use them according to the number of interpolation points. Note that the interpolation process may follow a known method, and the description thereof is omitted.
[0247]
In the next step S5, the initial learning is completed by registering the functional relationship obtained in the previous step as data in the ROM in one time. However, EPROM (erasable programmable ROM) or EEPROM (electrically erasable and writable) is completed. ROM) or the like can be used to easily cope with subsequent data correction, and the data can be stored in an auxiliary storage (or external storage) as a file as described above. It is.
[0248]
FIG. 40 is a flowchart showing a procedure for recognizing the shape from the pressure distribution information obtained in the state where the pressure detection sheet is wound around the object. First, in step S1, the object is inserted into the cylindrical pressure detection sheet. To do. For example, a pressure detection sheet is wound around the subject's arm.
[0249]
In the next step S2, after the pressure detection point is selected, pressure detection information is acquired by the pressure detection element corresponding to each point, and in the next step S3, the obtained information is compared with the ROM data.
[0250]
Then, in the next step S4, the shape data corresponding to the acquired pressure detection information is obtained from the ROM data. However, since there is no guarantee that the data that exactly matches is in the ROM, the ROM data close to the pressure detection information is searched. Based on this, predicted shape data is calculated. In order to specify the shape from the pressure information, it is important to determine whether or not recognition is possible and the accuracy of possessing a lot of data related to various shapes and making it into a database.
[0251]
When recognizing the surface shape of the human body using the pressure detection sheet (pressure detection means), it is a problem how to determine the joint, but the change in the surface pressure detected by the pressure detection sheet is large. It is preferable to adopt a configuration in which the range is recognized as a joint by the shape recognition means and the configuration data of the wire frame model as the shape model is output.
[0252]
41 to 44 are for explaining the shape recognition of the upper limb as an example.
[0253]
FIG. 41 is a flowchart showing an example of a shape recognition procedure, and it is assumed that the initial learning has been completed at the start time.
[0254]
First, in step S1, pressure information at each detection point is acquired in a state where the pressure detection sheet is wound over the entire upper limb surface.
[0255]
In the next step S2, the target person performs a certain operation so that the operation is stably performed. For this purpose, as shown in FIG. 42, a visual display device such as a head-mounted display (HMD) is mounted on the subject, and the movement of the upper limb to be imitated is projected on the device. In other words, as shown in the screen W in the figure, the target person operates by moving his / her hand following the virtual hand so that the image RH showing the actual hand overlaps the virtual image VH of the operation to be imitated. To imitate.
[0256]
In step S3, after selecting a region in which a pressure equal to or higher than a predetermined reference value is detected from each pressure detection point, a joint portion discrimination process is performed in next step S4. That is, as shown in FIG. 43, in the upper limb, the wrist, elbow, and shoulder have large changes in pressure due to the kinematics, so the joint can be recognized from the pressure information at the detection point corresponding to this portion. . In the figure, the position indicated by a circle of “+” conceptually indicates the joint, and the range indicated by the arrow “Ra” is the range where the pressure change is significant at the wrist, and the arrow “Rb”. The range shown is the range where the pressure change is remarkable at the elbow, and the range shown by the arrow “Rc” is the range where the pressure change is remarkable at the shoulder.
[0257]
In the next step S5, the length between the joints is obtained from the recognition result of each joint, thereby creating a bone wire model (representing the skeletal structure as a wire frame model) and acquiring shape data. The recognition result for the upper limb shape can be output as polygon data, for example. Moreover, although explanation is abbreviate | omitted, of course, it can follow the same method also about other site | parts, such as a leg.
[0258]
In recognizing the joint angle, when recognizing the joint part as described above, a point or range where a pressure change is newly applied to the operation of the even number is searched, and the joint angle is determined from the pressure detection information at the point or range. It is preferable to obtain. That is, as described above, the surface around the joint is a place where the pressure change due to the kinematic movement is remarkable, and the pressure gradually increases as compared to other places as the joint bending angle increases.
[0259]
In recognizing the even position, it is preferable to find a detection point that satisfies several pressure change patterns shown in FIG. 44 and obtain the position from the pressure change. In the graphs shown in FIGS. 44A to 44C, the horizontal axis is time t (the start point of the constant operation described in step S2 of FIG. 41 is the starting point of the time axis), and the vertical axis is detected. The time change is schematically shown by taking the pressure P.
[0260]
FIG. 44A shows a state where the pressure P increases with a substantially constant slope from the start of operation as indicated by the graph line g, and in FIG. 44B, as indicated by the graph line g1. In addition, after the pressure at a certain detection point first rises and saturates, the pressure at another detection point rises and saturates as shown by the graph line g2.
[0261]
In FIG. 44 (c), as shown by the graph lines gi (i = 1, 2,...) Indicating the respective pressures at a plurality of detection points, the pressures are saturated after rising one after another with a certain delay time. You can see how it goes.
[0262]
The position of the even number can be recognized by capturing a place where a characteristic pattern is recognized in such a temporal change in pressure, which is useful for recognition of hands, fingers, elbows, shoulders, and the like.
[0263]
Further, when recognizing the joint part whose shape is changed by the twisting operation by the shape recognition unit, it is preferable to obtain the joint angle from the pressure change caused by the twisting in the sheet-like base material constituting the pressure detection unit. For example, it is possible to recognize a shape change by capturing a muscle twist in the rotation of the neck and waist, the pronation / extraction of the forearm, and the like.
[0264]
Regarding the recognition of wrinkles formed due to the separation of the pressure detection sheet from the body surface due to the skin surface twisting, etc., a sudden pressure change continuously occurs over two or more detection points. It can be processed by regarding the place as a trap. However, when creating the shape data as polygon data, for example, the shape becomes complicated due to the inclusion of a large number of haze data, and the amount of data is inevitably increased. When it is desired to obtain only the surface shape data, a combination process relating to polygon data is used. This is a standard process used when compressing data by simplifying a polygon mesh and thinning out points and lines, taking into account how much it contributes to the importance and shape characteristics of edges. Various methods are known, such as a method that uses a volume change after edge removal or a distance sum between a new vertex set after edge removal and the vertex of a triangular patch as an evaluation amount.
[0265]
The above shape recognition method is effective for acquiring shape data of the skin surface of the human body. At this time, no imaging means for image recognition is required, and therefore, for example, the object described with reference to FIGS. It is also useful when acquiring the external shape data of a person. That is, the subject wears an effector in which the electrode part and the pressure detection part are formed in a multilayer structure on a sheet-like base material, and the shape data of the skin surface obtained by the pressure detection part at that time is acquired. If the external shape of the subject is recognized, a database showing the muscle arrangement can be created by specifying the body shape of the subject as described above.
[0266]
Next, the construction of the database using the above-mentioned item (II) information on the bone cross section, muscle and fat cross section will be described.
[0267]
So far, the method of acquiring data related to the outer shape and surface state of an object has been described with an interest, but helical CT (Computer Tomography), helical MRI (Magnetic Resonance Imaging), optical CT (Optical Computed Tomography): optical tomography Image measurement = one of the techniques for visualizing invisible information using a laser. For example, CT based on the projection theorem assuming that the forward multiple scattered rectilinear component detected by the optical heterodyne detection method follows the linear absorption law (Lambert-Beer law) When image information obtained by a tomographic imaging (or imaging) means such as an image reconstruction method using an algorithm is known, it is preferable to actively utilize these.
[0268]
In other words, after having the subject take a reference posture (for example, a basic standing posture) or fixing the target portion on the imaging table with a cast, the bone cross section obtained by helical CT and obtained by helical MRI, etc. A database of static human body structures related to the subject is constructed by acquiring image information of the sliced muscles and fat cross-sections and creating a database by stacking a large number of image data with different cross-sectional positions. And while acquiring the image data which imaged the kinematic pair along the time axis while operating the joint, and creating a static database concerning the shape of the kinematic pair for each angle of the joint, Build a human body structure database. Using this database to create a database that shows the muscle arrangement of the subject, it includes the internal structure, density distribution, and electrical resistance value distribution of bones, muscles, tendons, etc. in addition to the external shape of the subject (or target part). , You can get a more precise model.
[0269]
FIGS. 45 to 47 are flowcharts showing examples of such procedures.
[0270]
First, after selecting a kinematic pair to be imaged in step S1 of FIG. 45, in next step S2, the target part is fixed on the imaging stand with a cast in consideration of the state of the even number. Here, the “state” means a state in which a change in the joint angle is taken into consideration for a muscle that affects the motion of the even number. For example, with regard to the forearm, the state of the target region is set so that tomography can be performed in pronation and supination motion, taking into account movements related to finger measurement and actions such as crouching, buckling, and palmar bending. At that time, consideration should be given so as to distinguish between a state in which the muscles are tensed and a state in which the muscles are not tensed.
[0271]
In the next step S3, after setting the number of times of imaging in the longitudinal direction of the target region, the process proceeds to step S4, where bone imaging by helical CT and imaging of bones and muscles by helical MRI or optical CT are performed. That is, the bone contour can be identified using helical CT, and the muscle contour can be identified from the imaging data obtained by helical MRI or optical CT using a lubricating film between the muscles. . In the latter case (strain identification in a tomographic cross section), a method of automatically identifying by image processing using the color difference between the lubricant film and the stripe, and manual or semi-manual using image processing software There is a method of identifying by (for example, a partial auto-trace function related to contour extraction), but in any case, each muscle can be distinguished. As for fat identification, a known method for detecting a frequency difference, a phase shift, and the like has been established by utilizing the fact that there is a difference in resonance frequency between water protons (protons) and fat protons.
[0272]
In the next step S5, an area where the brightness difference (brightness difference between the white part and the black part) is remarkable is detected in the image information obtained by the helical CT. In other words, the bone region can be identified by extracting the contour where the black and white contrast is large because the contour portion of the bone is white and the periphery thereof is black.
[0273]
In step S6, a white region is extracted from the image information obtained by helical MRI or the like, and in the next step S7, data obtained by deleting the bone region is obtained by the difference calculation of the data obtained in steps S5 and S6.
[0274]
Then, in step S8 in FIG. 46, muscle data (including contour shape information) is extracted from the white region (the bone region has been removed in the previous step) in the image information obtained by helical MRI or the like. Assign the name of the line to this. In that case, a name is manually assigned to each extracted region, and a line is named automatically or semi-automatically (support for displaying candidate names, presenting options, etc.) using the model of the human body structure. There is a method.
[0275]
In the next step S9, after measuring and obtaining the length of the contour line obtained from the contour data of each bone or muscle, the process proceeds to step S10, where all the imaging operations in the longitudinal direction of the target part are completed. Judge whether or not. Then, the process proceeds to step S11 when the process is completed, but when the process is not completed, the remaining number of shootings is reduced by one, and then the process returns to step S4 in FIG.
[0276]
In step S11, after selecting a target from the bones and muscles obtained in the process up to the previous step, in the next step S12, the ratio of each contour line obtained in step S9 with respect to the image information in each tomographic section is determined. calculate. That is, a ratio is determined for the length of the contour length of the bone on another tomographic plane based on the contour length of the bone on a certain tomographic plane.
[0277]
For example, as shown in FIG. 48, the shape of the contour line on each tomographic plane “Sn” (n = 0, 1,...) Has a crescent shape, and the length of the contour line is set to “lgn”. When (n = 0, 1,...), In the figure, lg0 = 30, lg1 = 40, lg2 = 50, lg3 = 40, lg4 = 30 (the unit of length is omitted). Therefore, when lg0 = 30 is used as a reference, it is possible to obtain the ratio of each length with this being “100”.
[0278]
In the next step S12, the polygon point (the contour line data is planar data at this point in time, according to the ratio data (precisely the data group) obtained in the previous step). Thus, when generating three-dimensional polygon data, this point is used as the main vertex of the polygon (in this sense, it will be referred to as the “main polygon point”). This is referred to as “X0”). For example, the longer the contour length, the greater the number of points.
[0279]
In the example shown in FIG. 48, when the number of points for each contour length “lgn” (n = 0, 1,...) Is “X0_n” (n = 0, 1,...), X0_0 = 30, X0_1 = 40, X0_2 = 50, X0_3 = 40, and X0_4 = 30 (for convenience of explanation, the ratio of X0_n to lgn is set to “1: 1”).
[0280]
In step S13, a pointed portion (angular portion) in each cross section is identified, and the points (the number is “X1”) are extracted and registered as main polygon points.
[0281]
For example, in the crescent shape as shown in FIG. 49, there are sharp portions at both ends, and the vertices “MMP0” and “MMP1” are registered as the main polygon points, and in this case there are two X1s. It is.
[0282]
In step S14 of FIG. 46, the center of gravity of the portion (surface) surrounded by the contour line is calculated. For example, when a reference point “O” is set for a figure surrounded by a spline curve Sp as shown in FIG. 50, a large number of triangle groups (on the points P0 to Pn-1) ( When the target figure is approximated by Tr0 to Trn-1), the surface centroid (point G) can be obtained by combining the position vectors of the centroids (G0 to Gn-1) of the triangles. .
[0283]
In step S15 in FIG. 47, the contour line is divided at equal angles around the center of gravity G obtained in the previous step. In other words, when the number of divisions is “N”, “360 ° / N” straight lines (radial half straight lines) are drawn from the center of gravity G as “N = X0−X1”, and these and the contour lines (or splines thereof) are drawn. The intersection point with the curve) is obtained, and the intersection point is additionally registered as a main polygon point.
[0284]
FIG. 51 shows dividing lines Ln_0 to Ln_n-1 when the spline curve Sp (contour line) is divided into N equal parts (equal angle division) around the point G, and intersections Q0 to Qn-1 between these and the contour line. ing.
[0285]
In the case of FIG. 49, the surface centroid G calculated for the crescent shape (for example, the shape at the tomographic plane S0) and “N = X0_0−X1 = 30−2 = 28” lines passing through the centroid G. The figure is divided by the dividing line (center line) Li (i = 2, 3,..., 29) (see FIG. 5B), and each dividing line Li and the contour line (Sp) of the figure are separated. The intersection “MMPi” (i = 2, 3,...) Is calculated and registered as the main polygon point.
[0286]
In this way, the reason for subtracting X1 from X0 in the division number N is that points set at X1 pointed parts are registered in advance as main polygon points, so by reducing the number from X0 (assigned number) This is to make the number of dividing lines constant (by doing so, the ratio “X0_n / X0_0” (n = 1, 2,...) Of the allocation numbers for each tomographic plane is set so as not to be broken. Can do.)
[0287]
As described above, in creating the three-dimensional shape data from the contour lines of the bones or muscles in each tomographic section, after calculating the surface centroid of the inner region of the contour line, radially with equal angular intervals from the surface centroid When intersection points between the extended straight line group and the contour line are obtained and these intersection points are used as the reference points of the three-dimensional shape data, more detailed information content is obtained with respect to the points temporarily set in step S13 in FIG. Since points can be added, the degree of approximation to the contour line can be increased. In other words, it is preferable to reduce the data processing burden by accurately selecting the position of the point rather than simply performing higher-order approximation by increasing the number of points in the dark cloud (centering on the center of gravity G of the surface). Data processing is easier when polygon points are set to have equal angular intervals as shown by the above points Q1 to Qn, than the angular intervals of line segments connecting the center of gravity and the polygon points are not equal. Because it is.)
[0288]
Note that the number of dividing lines “N” is preferably increased in proportion to the size of the area from the viewpoint of high accuracy. However, in the above description, in determining N, the number of polygon vertices (X0) is used instead of the area. The processing time is reduced by substituting -X1).
[0289]
In step S18 of FIG. 47, it is determined whether or not there is a large disparity in the distance between the main polygon points (comparison determination as to whether or not the ratio of the distance between points or the average value or variance thereof is greater than or equal to the reference value. If the disparity is large, the process proceeds to step S19. If the disparity is within the allowable range, the process proceeds to step S20.
[0290]
In step S19, in order to correct the distance disparity between the main polygon points, an auxiliary polygon point (hereinafter referred to as “sub-polygon point” between the main polygon point where the disparity is a problem and the main polygon point adjacent thereto. "). If a main polygon point overlaps with another main polygon point or if the distance between the two points is smaller than the interval between adjacent sub-polygon points, the pre-registered main polygon point is given priority (N = X0). (Assuming that N = X0 instead of -X1, skip the overlapping points or the points with a narrow interval, and place the dividing lines.)
[0291]
Then, in the next step S20, it is determined whether or not a series of processes from steps S2 to S19 have been completed. If so, the process proceeds to the next step S21, and if not completed, the process returns to step S2 in FIG.
[0292]
In step S21, three-dimensional shape data (polygon data) is generated based on the main polygon point and the sub-polygon point, and the shape data is specified by the names of bones, muscles, and the like expressed by the three-dimensional shape data (polygon data). Generate.
[0293]
In addition, when expressing the movement of muscles with polygon data (dynamic expression of shape), a spline curve passing through the center of gravity (group of centroids) on each tomographic plane is defined as the center line, and the shape of the center line If the polygon data of the line is changed with time (according to the positional change of the polygon point), the shape of the line can be predicted.
[0294]
Further, the skin change can be expressed by defining a spline curve passing through the center of the surface of the even number. However, since it may be assumed that the subject is in an unloaded state at the time of imaging such as helical CT or MRI, the skin is preferably supplemented in consideration of the surface shape at the time of pressure detection. For this purpose, for example, the following procedure is taken.
[0295]
(1) Create a model with a skin model added to the skeleton and muscle model of the subject
(2) Acquire skin surface shape data by detecting surface pressure and compare this with the surface shape obtained from the skin model of (1).
(3) When the polygon data based on the skin model and the polygon data of the surface shape based on the pressure detection do not match within the allowable error range, it is determined which data is far from the center as viewed from the center of the polygon. (For example, the center of the even number is the reference point “O”, and the half line passing through the reference point intersects the polygon figure based on the skin model (polygon point) “PP1”, and the half line is used for pressure detection. (When the point that intersects the polygonal shape of the surface shape is “PP2”, the length of the line segment “OPP1” is compared with the length of the line segment “OPP2”, etc.)
(4) If the polygon data based on the skin model and the polygon data of the surface shape based on the pressure detection do not match within the allowable error range, the relative expansion of the area spread between the both is compressed or expanded. The ratio of the length of the line segment “OPP1” and the length of the line segment “OPP2” is calculated for an arbitrary point to obtain the compression ratio (or expansion ratio).
(5) Based on the ratio obtained in (4), the expansion and contraction processing is performed so that the sizes of the data of both models are substantially the same, and the reference center (the even number of images) created from helical CT, MRI, etc. Compression or expansion processing is performed on the center, etc. (this is an image size change, not data compression or expansion).
[0296]
In naming the bones, muscles, and skin, the center line is obtained from the data relating to the arrangement of the bones, muscles, etc. obtained from the model of the human body structure described above, and this is specified by the polygon data obtained in step S21 in FIG. It is preferable to specify the name related to the polygon data by checking the inclusion relation of whether or not it is included in the region (that is, the estimated region of bone or muscle), which enables automatic or semi-automatic identification processing become.
[0297]
Note that the number of times of photographing and the amount of data to be processed can be reduced by partially photographing the inside of the even number in the photographing in step S4 of FIG.
[0298]
In addition, when the quantitative ratio of the red stripe and the white stripe is obtained from the color information of the stripe distribution in the image information in step S8 of FIG. 46 (corresponding to the area ratio of each stripe on the tomographic plane, the distinction between red and white is It can be easily identified from the difference in color by image processing (applying a color filter).) The target reaction speed and endurance can be inferred from the ratio of red and white stripes (when the percentage of red stripes is high) Is excellent in endurance, and in the case of a large percentage of white stripes, it has excellent instantaneous power.)
[0299]
Next, the description will proceed to the (III) low frequency electrical stimulation method and interference wave electrical stimulation method.
[0300]
First, the low frequency electrical stimulation method will be described.
[0301]
Basic elements (stimulation conditions) in the low-frequency electrical stimulation method are as follows (refer to FIG. 52 for various quantities).
[0302]
i) Current intensity (intensity I)
ii) Stimulation pulse duration (pulse width Tw)
iii) Polarity and waveform of stimulation pulse
iv) Stimulation frequency
v) Stimulation time and rest time (Ts)
vi) Electrode arrangement
vii) Launch and mobilization of motor units by electrical stimulation.
[0303]
With respect to i) above, when a nerve or muscle is excited, in order to induce a potential change in the cell membrane, a certain level of current intensity and electrical energy are required. The stronger the intensity is, the higher the stimulation effect is, but also the discomfort and the pain become stronger. Therefore, in setting the stimulation intensity, it is necessary to consider the disadvantages of the effect, and 50 mA (milliampere) or less, for example, 20 mA. About 30 mA is a standard (with medical devices, an output current of 20 mA is the upper limit). It should also be noted that the stimulation intensity (or muscle contraction force) varies depending on the duration of ii) and the frequency of iv).
[0304]
If the (time) average value of the stimulation current is not zero, a polarization phenomenon occurs between the electrode and the living body, which causes an electrochemical composition change, which is not preferable. For this purpose, the direct current component is cut using a capacitor (capacitor), an insulating transformer, or the like.
[0305]
With regard to the duration of ii), the problem is how to define the duration of the stimulation pulse for exciting nerves and muscles by supplying a certain amount of electrical energy. In other words, the energy amount of the stimulation waveform is determined based on the product of the stimulation intensity I and the pulse width Tw. In general, as the pulse width increases, the degree of muscle contraction increases, but conversely, pain is induced. Therefore, a pulse width of about 0.2 to 0.3 milliseconds is used.
[0306]
It is known that there is a certain relationship between the stimulus intensity and the pulse width (so-called strength-time curve). When the stimulus intensity is weak, a long pulse width is required, and conversely When the intensity is high, muscle contraction is performed with a short pulse width. The intensity of the current when contraction does not occur below a certain current value no matter how long the pulse width is, is called “base current”, and the pulse width at twice the current value is Called value (chronaxie).
[0307]
For electrical stimulation applied to nerves and muscles, a pulse voltage (or current) having a short duration is used. For example, amplitude modulation (AM) is used as a modulation method for a stimulation pulse train for controlling the amount of muscle contraction. Pulse width modulation (PWM) and frequency modulation (FM) are used.
[0308]
With regard to the polarity of iii) above, the stimulation electrode is set as the cathode, and the indifferent electrode (electrode that does not feel stimulation when electrical stimulation is applied) is set as the anode. This is because the excitability of the muscle is generally high when the stimulation current that induces muscle contraction is under the cathode, and negative pulses are often used (for example, with negative rectangular waves, This reduces the discomfort as much as possible and produces a smooth and strong muscle contraction with less muscle fatigue.)
[0309]
In addition, in order to excite nerves and muscles, a steep current change of a certain level or more is required together with the amount of electrical energy, and the slope (or slope angle) at the rise of the waveform needs to be large to some extent. That is, even if a large amount of current is used, if the current intensity is increased very slowly, the stimulation does not result from adaptation. This “slope” (the angle at which the stimulation pulse rises, see “θ” in FIG. 52) is related to the stimulation waveform (90 ° for an ideal rectangular wave), and is negative for the waveform. Characteristic square wave, corrected single phase / exponentially increasing / decreasing wave, and the like.
[0310]
Regarding the frequency of iv) above, there is no clear definition since there are various ways of considering the range of “low frequency”, but a range of 200 Hz or less or 100 Hz or less is a guide. It should be noted that a single electrical stimulation only causes the muscles to undergo a single contraction, but does not cause a contraction. In other words, a stimulation frequency of a certain level or more is required to induce twitching, and normally it is about 15 Hz or more (muscle tremor appears when stimulated at a low frequency of 15 Hz or less). In addition, the contraction force increases as the stimulation frequency approaches the muscle reaction band frequency.
[0311]
The ratio between the stimulation time and the rest time in the above v) is a problem, and it is necessary to consider muscle fatigue when determining the duty cycle (or duty ratio). In other words, if the stimulation time is too long, muscle fatigue is easily caused. Therefore, depending on the stimulation intensity and frequency, a rest period of 1: 1 or more is generally required for the stimulation time. . If you want to increase muscle endurance, set the stimulation time to make the muscle contraction short while keeping the pause time shorter than the reference value, and if you want to increase the maximum muscle strength, while securing sufficient pause time. Such considerations are necessary.
[0312]
The electrode arrangement of vi) is preferably determined in consideration of the most responsive part (so-called motor point = motor point) when electrical stimulation is applied to induce muscle contraction. That is, after recognizing the shape and movement of the target portion by the shape recognizing unit 1c described above, the muscle arrangement is specified and the muscle stimulation region is determined to include the motion point. The movement point is a part where the nerve enters the muscle, and the place is considered to be near the neuromuscular junction.
[0313]
In addition, when the nerve trunk is superficially under the skin, the site also reacts well. Therefore, for the electrode arrangement and shape to be applied to the stimulation to this movement point, the energization method is a monopolar method, bipolar method It is preferable to stipulate separately by law and group stimulation method.
[0314]
That is, in the monopolar method, stimulation is performed by placing a large indifferent electrode at a site with few muscles away from the target muscle, and placing a small stimulation electrode (related electrode) on the motion point. For example, in the effector described above, an indifferent electrode having a large occupied area may be configured by a large number of electrodes, and an indifferent electrode having a small occupied area may be configured by a small number of electrodes.
[0315]
In the bipolar method, stimulation is performed by placing the motion point between relatively large electrodes of the same size or by placing the cathode at the motion point. In the effector described above, the stimulation area can be easily set by selecting the number and position of the electrodes used.
[0316]
The group stimulation method uses an electrode arrangement that contracts an entire muscle group.
[0317]
In the case of low frequency electrical stimulation, when multiple pairs of electrodes with one pair of two poles are installed at the target site, the electrodes are arranged so that the predicted conduction lines (predicted lines of the stimulation current path) between the electrodes of each set do not cross each other. It is desirable to set This is because there is a possibility that a strong interference wave stimulus may be applied to the streak at the place where the conduction prediction line intersects, and it is necessary to avoid such a situation. In the calculation and determination of the predicted conduction line between the electrodes, the resistance component of the adjacent electrode in the electrode group is detected, and the internal structure data of the target part obtained from the above-described tomographic plane information is bone, muscle, It can be calculated by setting the resistivity (or specific resistance) according to the fat material. For example, when the electrodes 2, 2,... Are arranged in a grid as described above, there are eight adjacent electrodes around the selected electrode, but these electrodes By selecting one of the electrodes and the center electrode and applying a certain weak voltage, the resistance value between the two can be known, and the resistance value distribution on the surface can be obtained by performing such detection for the other electrodes as well. Can know. In addition, regarding the resistance distribution related to the internal structure, the known resistivity related to each component (bone, muscle, tendon, fat, etc.) is set in each part based on the result of the tomographic analysis described above, and then the finite element (A conduction prediction line is calculated as a path passing through a place having the smallest resistance between both electrodes when a paired electrode is specified).
[0318]
With regard to vii), in normal voluntary muscle contraction, the motor unit that fires as the load increases increases from the slow muscle (type I) muscle fiber dominated by the thinner nerve fiber to the thicker nerve fiber Participates as firing of fast muscle (type II) muscle fibers (this is called “Size principal”).
[0319]
However, in contrast to electrical stimulation, if the stimulation intensity is increased, muscle fatigue will appear early because stimulation is performed from thicker nerve fibers. It is necessary to set stimulation conditions.
[0320]
In addition, in the case of electrical stimulation, stimulation is performed from thick nerve fibers (that is, motor nerve fibers), and finer sensory nerve fibers are subsequently stimulated, so that the stimulation threshold for motor nerves is within the normal range. For example, it is possible to obtain muscle contraction by stimulating motor nerves without giving pain due to superficial sensory nerve stimulation.
[0321]
Next, the interference wave electrical stimulation method will be described.
[0322]
This method generates a low frequency current (beat current) corresponding to a frequency difference using a medium frequency current (for example, 4000 Hz and 3900 Hz) that can ignore skin resistance. There is an advantage that sufficient electric stimulation energy can be supplied to the target part because it is a method that generates a low frequency in a part of the internal tissue of the human body (when a low frequency current is directly applied to the skin) It is difficult to supply enough energy to the internal tissue because most of the electrical energy is consumed by the electrical resistance of the skin.) Further, interference wave electrical stimulation is desirable for inner layer muscles (for example, finger extensor muscles) in which muscle contraction cannot be visually recognized from the skin surface.
[0323]
The basic elements (stimulation conditions) of interference wave electrical stimulation are as follows.
[0324]
i) Stimulus frequency difference (interference frequency)
ii) Energy amount of beat current.
[0325]
As stimulation conditions suitable for muscle stimulation, for example, when a solid conductor wave (tripolar interference wave) is used as a pulse waveform, the output current is 100 milliamperes or less, the carrier frequency is 5000 Hz, and the interference frequency is about 0.5 to 200 Hz. . As stimulation conditions suitable for skin stimulation, for example, when the pulse waveform is a negative rectangular wave, the output current is 50 mA or less, the carrier frequency is 4000 to 4150 Hz, and the interference frequency is 150 Hz or less.
[0326]
In setting the electrodes used in the effector, there is a method of experimentally examining the state of interference wave stimulation. However, an interference wave simulator (an electromagnetic field simulator used for designing mobile phone antennas, etc. It is preferable to use a device for simulating interference waves related to a plurality of motion points by determining different propagation coefficients for the like.
[0327]
In that case, if the attenuation rate (σ) is set for each material for bone, muscle, fat, and tendon, and the electrode is set in a place where the energy loss to each part of the skin surface is small, efficiency in terms of energy consumption Is. That is, when internal tissue information (that is, bones, muscles, etc.) is obtained from the tomographic plane information of the target site by the above-described method, attenuation for each material with respect to the three-dimensional shape data (polygon data). Interference wave simulation can be performed by setting the rate.
[0328]
In FIG. 53, an arbitrary point Sa (Sax, Say) on the skin surface S and the inner point P of the part XP such as a bone or muscle are separated by a distance “r”, and the wave attenuation rate with respect to the part XP This indicates that the energy loss “L” when “σ” is given is expressed as a function L (Sax, Say) of the position coordinates of the point Sa (in the figure, simply “L (Sa)”). ). That is, by detecting a place having a small function value for the inner point P by simulation, the electrode located at the place or the electrode closest to the place can be set as the use electrode. When the electrode to be used is determined, the inner point P of the target portion at that time becomes an interference point, and it is desirable to set this point for each joint drive angle, and the muscles that are dense like the finger muscle group. Regarding the arrangement, it is desirable to set the interference points at different positions in the longitudinal direction of the even number (to prevent the interference points from being concentrated in a narrow area).
[0329]
Furthermore, if the substance density or mass density (ρ) is set for each part XP for each part XP, and the electrode is set in a place where the wave phase shift to each part is small on the skin surface, for the interference wave stimulation An effective location can be selected. The substance density (ρ) can be determined from the mass density (mass per unit volume) corresponding to the material.
[0330]
And in order to acquire the ideal interference wave at the inner point P of the part XP, it is preferable to take the following procedures (1) to (4) for setting the electrode position.
[0331]
(1) Wave reflectivity (α) and refractive index (β) are set for each material of bone, muscle, fat and tendon.
(2) As described above, the electrode at the location of the minimum energy loss and the location of the minimum phase shift based on the attenuation rate (σ) and the material density (ρ) set for bone, muscle, fat and tendon, respectively. A plurality of electrodes are selected from the candidates (FIG. 54 shows the electrode candidate positions on the surface S by three points “Sa1”, “Sa2”, and “Sa3”).
(3) By simulating the waveform reflected or refracted by the part XP with respect to the temporal transition of the carrier wave from the electrode position selected in (2) to the inner point P of the target part XP, The frequency distribution of interference waves in
(4) The steps (2) and (3) are repeated until an ideal interference wave at the inner point P is obtained, and finally the electrode position to be used is determined.
[0332]
In the procedure of (2) above, rather than selecting electrode candidates considering the attenuation rate (σ) and the material density (ρ) from the beginning, in the initial trial, the attenuation rate (σ) and the material density (ρ) After selecting the electrode candidates without considering the electrode, it is more efficient to set the attenuation rate (σ) and the material density (ρ) and select the electrode candidates based on this after the prospect of the electrode location has reached a certain level Preferred for.
[0333]
In addition, it is preferable to configure the apparatus so that the output of the electrical stimulation to the inner point P of the target site XP can be set by changing the output current value of the electrode set in this way (electrode in the interference wave) This is to make an appropriate output setting for the arrangement.) In addition, the muscle placement region sometimes changes significantly depending on the joint movement (for example, forearm pronation / extraction), so a simulation is performed for each joint movement (that is, the dynamic It is preferable to determine the position of the electrode to be used by performing an interference wave simulation that considers the passage of time based on a human body structure database (only a static interference wave simulation that does not include the passage of time of the target area) Because it is difficult to set the electrodes for the interference waves to the corresponding muscles properly.) At that time, a database that associates information indicating the kinematic state of the kinematic pair (including polygon data, etc.) with information on the ideal stimulation region and the electrode position to be used is constructed, and the data is regenerated. It is desirable to use it (for example, to give priority to each electrode used, or to include attribute data in the database for substituting with another candidate electrode when a failure occurs in the electrode used). For this purpose, after acquiring a plurality of electrode candidates, the position information is stored in a storage unit such as a memory or an auxiliary storage device, and when actually used as an electrode, the position information of the electrode candidates is read out and these position information is read out. Select an electrode suitable for the target area of electrical stimulation.
[0334]
In the above description, the object of simulation is the inner point P. However, in the electrical stimulation to the inner layer muscle group, the simulation is performed on the surface in the region having a constant (width) frequency band instead of the point, Furthermore, there is a method of performing a simulation on a solid (for example, a sphere, a cylinder, a polyhedron, a prism, etc.) by expanding the surface, and the processing becomes more complicated as the number of dimensions increases.
[0335]
In the former method, in FIG. 53 and FIG. 54, when a plane is set in the interior or boundary of the target site XP instead of the point P, interference is performed by simultaneously stimulating a plurality of muscles or increasing the area of the region that responds to electrical stimulation. Wave simulation can be realized.
[0336]
In the surface setting, there are the following cases.
[0337]
a) Simultaneous stimulation to adjacent muscles
b) When a muscle that you do not want to move and a muscle that you want to move are mixed.
[0338]
First, in a), if the target surface is set so that adjacent muscles among a plurality of muscles are stimulated simultaneously, a simulation result of interference waves to a specific muscle and a nearby muscle can be obtained simultaneously. For example, for simultaneous operation of multiple adjacent muscles, electrical stimulation is performed over a wide range of interference areas so that the ideal frequency band region for each center of adjacent muscles is included in the multiple muscle ranges. It is preferable to carry out. When an actual electrical stimulus is applied, the surface set at the time of the interference wave simulation is set as a stimulation region, and an electrode to be used is selected from electrode candidates suitable for the stimulation region, and the electrical stimulation is applied.
[0339]
Further, in b), it is possible to exclude, from the stimulation target, a muscle that is not desired to be operated by properly using, for each region, whether the inner point P or the surface is set for the target part XP. In other words, when the muscles to be acted on and the muscles to be deactivated are intermingled, interference wave simulation is performed by setting a plurality of point-like or planar stimulation regions for these muscles. It is preferable.
[0340]
When selecting an electrode for electrical stimulation to the inner layer muscle group, when performing an interference wave simulation using a stimulation region having a fixed (width) frequency band as a three-dimensional region, that is, when performing simulation on a three-dimensional region in the target site XP For example, when a plurality of surfaces are designated as “SPi” (i = 1, 2,...), A boundary surface determined as “ΣSPi” (Σ represents the sum of i) and an arbitrary area in the inner region thereof Since the solid region is determined by the point and the interference wave simulation result is obtained, the region (volume) that responds to electrical stimulation can be made larger than that of the surface.
[0341]
In addition, when simultaneously simulating the low-frequency electrical stimulation and the interference wave electrical stimulation, the simulation processing for the interference region is performed after assuming the expected region of the low-frequency stimulation, so that the low-frequency electrical stimulation and the interference wave electrical stimulation are simultaneously performed for the target region. In addition, it is possible to select an appropriate electrode position for a stimulation region in the case where stimulation by an interference wave is applied (that is, it is necessary to avoid crossing of conduction prediction lines as described above for low-frequency electrical stimulation).
[0342]
The amount of calculation necessary for the interference wave simulation as described above becomes more conspicuous as the number of candidate electrodes increases. Therefore, in order to reduce the burden on the simulator and improve the processing speed, the following procedure is taken.
[0343]
(1) Use obtained based on the angle of the joint to be driven and the three-dimensional shape data (polygon data of bone, muscle, fat, tendon, etc.) for each joint angle and the positional relationship of each part based on the data Associating electrode position candidates and constructing them in advance as a database
(2) For a certain joint angle, electrical stimulation is actually performed at the position of the electrode used as a candidate in (1).
(3) The driving amount of the even number obtained by the electrical stimulation of (2) and the driving amount of the even number corresponding to the joint angle of (1) are compared to determine whether or not they match.
(4) The steps (2) and (3) are repeated while changing the joint angle.
[0344]
If no coincidence of driving amounts is recognized in step (3), information necessary for correcting the difference (for example, a difference in joint angle) is registered as data, or the position of the electrode used and the stimulation intensity are used. It is necessary to perform error compensation by changing the. That is, when the driving of the kinematic pair does not go as expected, there are a method of changing the set value of the stimulus output to the electrode, a method of selecting another electrode, and a method of using both.
[0345]
In addition, simplification of polygon data is also effective for simulation. For example, for a muscle, a wire frame model in which the centroids of polygon cross-sections are combined is adopted (for example, three-dimensional polygon data indicating a muscle shape is converted into a wire frame). A database is created after simplifying the data.), And various parameters are uniformly defined for each muscle.
[0346]
When operating multiple muscles simultaneously, in addition to applying electrical stimulation to these muscles at the same time, apply electrical stimulation to the muscles to be contracted in a time-sharing manner within the time when the muscles contract and do not expand again. There is a method of applying (a method of setting a point-like, planar, or three-dimensional region as an interference wave stimulation region and applying electrical stimulation to these regions by time division processing). The latter method is also advantageous for preventing low-temperature burns in that it does not continuously apply electrical stimulation to the same part of the skin.
[0347]
When the electrical stimulation is actually applied to the muscles, interference waves are generated with the interference wave area in the form of dots, planes, or solids. In this case, the interference wave areas are set so that the mutual interference wave areas do not interfere with each other (because there is a possibility that the independent operation of the muscles cannot be guaranteed if the wrinkle areas overlap). In addition, since the effects of the gravity on the even number and the joint cannot be ignored, it is preferable to stimulate the muscle in consideration of the relationship between the position of the joint and the even number and the gravity applied thereto (for example, the joint) When a certain position or posture is taken, the contraction ratio between the extensor and flexor muscles is changed according to how much weight is added to them and at what angle. Inconveniences such as unstable operation of the even number can be prevented due to the influence.)
[0348]
Next, the description will shift to (IV) Phantom Sensation.
[0349]
In this case, electrical stimulations having different frequencies are applied to the skin of the surface layer from the plurality of electrodes, thereby providing a tactile sensation. In other words, by applying the above-described interference wave electrical stimulation to the skin surface, it becomes possible to present tactile stimulation to the subject.
[0350]
In that case, the following presentation is mentioned.
[0351]
i) Point touch
ii) Line tactile sense
iii) Touch sense
iv) Change from point touch to surface touch or vice versa
v) A change in the stimulation region due to movement of a point tactile sense or a line tactile sense.
[0352]
First, i) presentation of a point tactile sensation can be realized by expressing the interference wave region as a point (interference point). That is, by giving the interference wave electrical stimulation by designating the inner point P of the target part, it is possible to obtain a touch as if the skin surface was pressed with a pointed object.
[0353]
Ii) Line tactile sensation can be realized by expressing the interference wave region as a line (that is, a line connecting a plurality of interference points). In this case, a feeling as if the skin surface was pressed by the side surface of the plate can be obtained.
[0354]
iii) The surface tactile sensation can be realized by expressing the interference wave region as a surface. In this case, a pressure sensation as if the skin surface was pressed by a flat plate or a curved plate is obtained.
[0355]
Regarding ii) and iii), focusing on the fact that the line is a set of points and the surface is a set of lines, the point-like (or linear) interference wave region is moved at high speed (the subject's It is necessary to move the region within such a short time that the user is not aware of the movement of points and lines.
[0356]
For iv), the electrical stimulation is applied by changing the interference wave region in time from the point to the surface or vice versa. For example, in order to achieve a feeling of landing when receiving an attack from an enemy that appears visually in a virtual space when applied to a game machine or the like using virtual reality, Then, the electrode position and range may be selected so that the stimulation area is converged from a circle to a point, and then electrical stimulation may be given to the subject.
[0357]
For v), it is a touch accompanying a change in the stimulation area when the point touch or the line touch is slowly moved. For example, in application to a game machine using virtual reality, in order to realize a feel when a sword is injured by an attack from an enemy visually appearing in the virtual space, a dot or line It is only necessary to apply electrical stimulation to the subject while changing the electrode position or range in time by moving the shaped stimulation area along a line along the sword (showing the trajectory of the sword).
[0358]
In the case of simultaneous simulation of interference wave and phantom sensation, the interference area is assumed after assuming the stimulation area of the interference wave related to the part to which stimulation is applied by phantom sensation and the stimulation area to the inner muscle by the interference wave. By performing the simulation calculation process, it is possible to select an appropriate electrode position for the stimulation region when the simultaneous stimulation is applied to the target part.
[0359]
Next, (V) a motion prediction method using an electromyogram will be described.
[0360]
Conventionally, an electrotherapy device using EMG feedback is known, but muscle tonus (muscle tonus) is a faint resistance felt by the subject when the muscle is stretched passively, The essence is muscle stretch reflex, but it is complexly controlled by the central nerve, peripheral nerves (α, γ-system), and muscle itself, and muscle stretch reflex means that when a muscle is passively stretched, Is a reflex that produces a force as a result of contraction due to a motion command from the spinal cord), and appears in EMG along with the motion set signal of multi-muscle fibers by ATP (adenosine triphosphate), so it works before muscle movement The prediction could not be made. The “multi-muscle fiber motion set signal” is generated along with the contraction of muscle cells by muscle tonus due to nerve excitement when a command is transmitted from the central nerve to the terminal nerve. It varies depending on the physique, blood glucose level, and muscle fatigue.
[0361]
Therefore, the muscle tonus is predicted as follows, and the motion is predicted by learning the relative relationship between the predicted muscle tonus and the surface electromyogram.
[0362]
FIG. 55 shows an example of an apparatus for acquiring a motion set signal of a multi-muscle fiber by ATP. In this case, the target site is a biceps brachii muscle (a biarticular muscle, a bending action related to the elbow joint and a rotation related to the forearm. It has an external action.
[0363]
A negative (polar) electrode 20N and a detection electrode 20D are arranged on the muscular abdomen of the biceps brachii muscle, and a positive (polar) electrode 20P is arranged on the muscle start point, and an electric stimulation generator (for example, An electrical stimulation signal from a medical low-frequency treatment device 21) is supplied.
[0364]
A two-channel FFT (Fast Fourier Transform) arranger 22 prepared for frequency analysis of the myoelectric waveform is supplied to the first channel (channel) via the amplifier 23 after the supply signal to the positive electrode 20P branches at point A. To the FFT unit 22a. Further, a signal extracted from the detection electrode 20D via the amplifier 24 is sent to the FFT unit 22b of the second channel, and the output of each FFT unit is sent to the differential operation processing unit 22c in the subsequent stage, and its output Is taken into the computer 25 after digitization. Note that a control signal necessary for timing control or the like is sent from the computer 25 to the FFT organizer 22.
[0365]
The frequency component of the signal sent from the electrical stimulation generator 21 to the positive electrode 20P is analyzed by the FFT unit 22a, but this is an original signal of electrical stimulation and does not include a muscle response.
[0366]
The frequency component generated by the electrical stimulation to the muscle is analyzed by the FFT unit 22b from the detection electrode 20D via the amplifier 24. That is, frequency analysis is performed on the response generated by following the muscle to electrical stimulation. The time required for the FFT process (ΔtF) must be sufficiently shorter (ΔtF <ΔtM) than the time (ΔtM) required for the subsequent movement change after the occurrence of muscle tonus of the subject (∵ “ΔtF> ΔtM). ”, The frequency analysis process cannot catch up with the movement of the muscle, so it becomes impossible to predict before the movement).
[0367]
The difference data obtained as a result of the difference calculation by comparing the frequency component for each band in the difference calculation processing unit 22c is the frequency component generated from the muscle by the electrical stimulation by the electrical stimulation generator 21 (excluding the frequency component of the original signal). ). That is, when the muscle contracts only by applying electrical stimulation, the potential information associated with the contraction of the muscle cells can be acquired without generating the muscle tonus. Therefore, the difference from the EMG data including the contraction of the muscle tonus and the muscle cells can be obtained. By subtracting the data (multiple muscle fiber motion set signal by ATP), muscle tonus can be predicted, and the muscle characteristics of the individual of the subject can be acquired.
[0368]
FIG. 56 is a graph schematically showing the frequency characteristics after FFT, and the electromyogram frequency characteristic chart is shown with the frequency (f) on the horizontal axis and the power spectrum on the vertical axis.
[0369]
The graph curve ga in the figure shows the FFT analysis result of the signal including muscle tonus and muscle cell contraction when the muscle is moved at the subject's will, and the graph curve gb is when the electrical stimulation is applied to the muscle. The frequency component of the myocyte contraction obtained from the above difference data is shown. That is, the part (see the hatched line in the figure) obtained by subtracting the component of the graph curve gb from the frequency component shown in the graph curve ga represents the component related to muscle tonus.
[0370]
Therefore, by storing and learning by associating the muscle tonus predicted based on the above difference data when various waveforms of the electrical stimulation applied to the muscle are changed and the potential information of the surface electromyogram at that time The operation can be predicted.
[0371]
The above-described myoelectric potential information necessary for prediction and the prediction procedure using the information are summarized as items in the following manner.
·Necessary information
A) Myoelectric potential information obtained when the subject operates the muscle according to his / her will (this will be referred to as “first myoelectric potential information”).
B) Myoelectric potential information obtained when the muscle is operated only by electrical stimulation (this will be referred to as “second myoelectric potential information”).
・ Prediction procedure
(1) The electrical stimulation is obtained by acquiring the analysis result of the frequency component of the electrical stimulation signal to the muscle and the analysis result of the frequency component of the signal including the frequency component generated from the muscle by the electrical stimulation and comparing them. Thus, only the frequency component generated from the muscle (that is, the waveform component that accompanies the contraction of the muscle cell not depending on the muscle tonus) is extracted as difference data. That is, since the frequency component related to the second myoelectric potential information includes the frequency component of the electrical stimulation signal, the component of the motion set signal of the multi-muscle fiber by ATP is obtained by obtaining the component from which the frequency component is removed. Can be extracted. This corresponds to a component reflecting the characteristics of the motor unit during muscle contraction.
[0372]
(2) Acquire components related to muscle tonus from the difference data obtained in (1) and data of myoelectric waveforms (including muscle tonus) obtained when the subject tries to move muscles on his / her will To do. That is, the frequency component related to the first myoelectric potential information includes the component related to muscle tonus and the component of the motion set signal of the multi-muscle fiber by ATP. By subtracting from one myoelectric potential information, only frequency component data relating to muscle tonus is obtained and used as muscle tonus prediction data. This is because the frequency characteristics obtained by FFT analysis indicate that “the EMG data (frequency component) as first myoelectric potential information (frequency component thereof) -multiple muscle fiber motion set signal (frequency component) of ATP” = muscle tonus (frequency of This corresponds to the conceptual diagram of “component”).
[0373]
(3) By repeating various steps (1) and (2) with various changes in the electrical stimulation signal waveform, the relationship between the predicted muscle tonus and the surface myoelectric waveform data is memorized and learned (for example, neural network)・ Network usage, etc.) That is, the muscle tonus prediction data calculated on the basis of the first myoelectric potential information and the second myoelectric potential information is used as the motion pattern of the subject (the time-dependent change in the kinematic pair motion accompanying the contraction of the target muscle). This is a series of data, and in addition to polygon data indicating the shape and parameters indicating the shape change such as the joint angle, the shape and motion such as hand adduction, abduction, and pronation / extraction of the arm are specified. Data, etc.) for storage and learning.
[0374]
(4) When the subject obtains myoelectric waveform data when he / she intends to move the muscle, he / she predicts the muscle movement from the muscle tonus predicted based on the information acquired in (3). . That is, frequency analysis is performed on the myoelectric potential information when the subject intends to operate the muscles, and the muscle tonus component data calculated from the result is acquired. Then, the data and the prediction data that has already been stored and learned are compared and collated, and when a match between the two data is recognized, a motion pattern of the muscle corresponding to the data is predicted. In this way, it is possible to predict muscle movement by knowing muscle tonus before the muscle actually moves (muscle tonus data includes frequency components having different characteristics for each movement. You can see the action that is going to grab a heavy object right before that.)
[0375]
As described above, muscle tonus can be predicted based on information obtained by removing the frequency component of the motion aggregate state signal of the multi-muscle fiber by ATP from the frequency component related to the first myoelectric potential information detected from the electrode. (This is handled by the control means 1d.)
[0376]
If the difference data obtained by the FFT units 22a and 22b corresponding to the frequency analysis unit 1f can be acquired, the state of muscle cell contraction when receiving electrical stimulation can be understood. Data relating to muscle characteristics can be acquired. For example, muscle characteristics include differences in muscle reaction speed and endurance due to the ratio of red muscle to white muscle, but myoelectric potential information associated with muscle cell contraction includes physical condition, blood glucose level, muscle fatigue. (In other words, the fact that this component is not constant depending on the state of the muscle is the cause of the difficulty in predicting the movement of the muscle, and the present invention is based on the first myoelectric potential information. (The focus is on predicting muscle movements without considering the characteristics of the motor units by performing processing to remove the components.)
[0377]
55, the electrode of the electrical stimulation generator 21 is described as an example. However, in the effector described above, the electrode position suitable for electrical stimulation (that is, the position selected by the interference wave simulator) is the same position. Muscle tonus can be predicted by acquiring myoelectric waveform data from the electrodes and performing frequency analysis on the signals from the anode side electrode and the cathode side electrode. That is, in addition to applying electrical stimulation to the muscles through the electrodes 2, 2, ... of the effector, these electrodes can also be used for acquiring electromyogram information (that is, as an electromyograph) Available.).
[0378]
FIG. 57 conceptually shows a configuration example in the case of performing electrical stimulation through the electrodes by switching control in the input / output interface unit (input / output selection processing unit 1e) and acquiring myoelectric potential information from the electrodes by time-sharing processing. (In the figure, an equivalent circuit for one electrode is shown).
[0379]
The signal supplied from the electrical stimulation generator 1b passes through the output buffer 26 and is then sent from the switching unit 27 (indicated by a switch symbol in the figure) to the electrode unit 1ad. In addition, a signal that has passed through the switching unit 27 from the electrode unit 1ad is sent to the myoelectric potential information acquisition unit 29 (including the FFT units 22a and 22b and the difference calculation processing unit 22c) via the input buffer 28. The state of the switching unit 27 is defined by a control signal sent from a control unit (not shown). When the output buffer 26 side is selected in the switching unit 27, electrical stimulation is applied to the muscle through the electrodes, and the input buffer 28 side is When selected, information on the electromyogram is acquired through the electrodes.
[0380]
In this way, by switching input / output (I / O) by time-sharing processing, it is possible to perform application of electrical stimulation and acquisition of myoelectric potential information through the same electrode, or the same for both Use different electrodes so that the application of electrical stimulation through the selected electrode and the acquisition of myoelectric potential information using the electrode as the detection electrode are not performed on the same electrode at the same time. Need to choose).
[0381]
If the actual motion obtained when applying electrical stimulation to the muscle through the electrode used selected by the above method is different from the result of the interference wave simulation, the position of the electrode used is changed to another candidate electrode (candidate registered). It is preferable to select an electrode position that best matches the actual operation by changing to the position of the electrode.
[0382]
Thus, by grasping in advance the muscle movement from the muscle tonus predicted from the frequency analysis and using the dynamic human body structure database, the next motion of the subject's joint can be predicted. That is, by using the database for associating the predicted muscle tonus and the surface electromyogram waveform data in the above procedure (4) and the above-described dynamic human body structure database, the motion (motion) predicted next for the target region. It becomes possible to learn the information on the basis of the surface EMG waveform data.
[0383]
It is also possible to use electromyograms when acquiring muscle work volume and even kinematic speed when muscle contraction occurs without moving the pair even when the wall is pushed by hand (stationary contraction). Preferably, for example, after having the subject wear a visual display device such as a head-mounted display, the operation projected on the device is imitated, or the joint is temporarily fixed with a brace or a certain load is applied. Get data about myoelectric pattern while applying.
[0384]
Next, (VI) prevention of low-temperature burn and adjustment of stimulation output will be described.
[0385]
In order to prevent low-temperature burns due to insulation contact resistance, the following methods can be mentioned.
[0386]
a) A method of changing the used electrode by limiting the use time so that electrical stimulation is not applied using the same electrode for a certain period of time or more
b) A method in which the stimulation intensity from the electrode is reduced or the use of the electrode is prohibited when the insulation resistance of the skin surface is greater than or equal to a predetermined threshold.
[0387]
First, regarding the change of the electrode used in a), there is a method in which an electrode at a position different from the electrode is used as an alternative electrode when the resistance value between the electrodes to be used is equal to or greater than a threshold value. For example, a method by changing the arrangement pattern of electrodes, a method of limiting the duration of energization time to each electrode by time-sharing processing (for example, belonging to each group so that adjacent electrode groups are not used simultaneously) When the electrode group is used in accordance with the time-sharing process, it is possible to avoid applying a stimulus for a long time to the same region of the skin surface.
[0388]
B) measures the resistance value between the electrode groups to be used, and if it is determined that the resistance value is equal to or greater than the threshold value, reduces the intensity of the electrical stimulation applied through the electrode. Alternatively, the use of the electrode is forcibly stopped. In addition, regarding the strength reduction, an allowable range for stimulation is taken into consideration after grasping the body fat percentage of the subject and the individuality of pain. In addition, regarding the prohibition of use of the electrode, temporary prohibition (prohibition with a release condition. For example, measure the insulation resistance value when the power is turned on again, and if this is below the threshold, cancel the prohibition of use, etc.) Prohibition (prohibition without setting a release condition). For electrodes that have been declared banned, either eliminate this and perform interference wave simulation again to select candidate electrodes and find alternative electrodes for prohibited electrodes, or give priority to the already registered candidate electrodes. It is necessary to select a substitute electrode for the forbidden electrode in consideration of the degree and the like, and it is possible to guarantee a safe use electrode. In addition, low temperature burns are likely to occur when skin resistance increases due to adhesion of dirt, dirt, etc., so it is possible to prohibit electrical stimulation when it is detected that the insulation resistance of the skin exceeds the allowable range. is necessary.
[0389]
Since the human fat directly becomes the insulation resistance in electrical stimulation, it is deeply involved in the discomfort caused by stimulation (that is, low-temperature burns are caused by Joule heat proportional to the resistance value).
[0390]
Therefore, it is preferable to classify the subject according to their body fat percentage, and to perform subjective evaluation at a finite stage (for example, five stages including comfort and discomfort) for the stimulus intensity and discomfort for each class (∵ Because the body fat rate alone is not enough information on how to get fat.) By using this result, it is possible to identify the individual difference regarding the low temperature burn and adjust the output of the electrical stimulation so as not to exceed the maximum stimulation value allowed for the subject.
[0390]
58 and 59 are flowcharts showing an example of a subjective evaluation method based on human body fat and stimulation. In this example, the evaluation target region is the biceps brachii, which is a flexion muscle, and the evaluation of the stimulation intensity is performed on both arms of the subject. (For example, in the state where both arms are straightly extended forward, the greater the discomfort, the greater the distance between the hands.) This method uses a method for recognizing the relative recognition of individual subjects (because it is difficult to evaluate absolute discomfort by using an absolute index).
[0392]
First, in step S1 of FIG. 58, a subject is allowed to wear the above-described effector on the upper arm, and then a plurality of electrical stimuli (for example, three stages of stimuli A, B, and C) are individually transmitted through the effector. After giving, in S2, in the next step, the subject is verbally explained so that the subject can recognize the stimulus A as the minimum, the stimulus B as the middle, and the stimulus C as the maximum stimulus.
[0393]
In the next step S3, the degree of discomfort of the stimulus C to the subject is expressed by the distance between both arms, and then the process proceeds to step S4 to select the stimuli A, B, C randomly (randomly). To the subject. The number of times is a predetermined number (for example, about 15 times).
[0394]
In the next step S5, the subject is asked how much discomfort each stimulus (A, B, C) gives by the distance between both arms. Thus, the evaluation is performed based on the relative value based on the distance between the arms in the case of the stimulus C (for example, the interval in the case of the stimulus C is 100, the stimulus B is 50%, the stimulus A is 25%, etc. Etc.)
[0395]
Next, in step S6, it is asked whether or not the random stimulus application has been completed for a predetermined number of times. When finished, the process proceeds to step S7 in FIG. 59, but when not completed, the process returns to step S4.
[0396]
In step S7, the intensity of the stimulus C is increased by one step from the previous level and applied to the subject. In the next step S8, the subject expresses how much discomfort the stimulus gives at the interval between both arms.
[0397]
Then, in step S9, it is determined whether or not both arms of the subject are opened more than a predetermined interval. If so, the process proceeds to the next step S10, and if not, the process returns to step S7.
[0398]
In step S10, a place where the change is remarkable is identified based on the myoelectric potential information obtained from the electrode group of the effector, and data of a predetermined number of electrode positions and myoelectric patterns are obtained from the higher order.
[0399]
In this way, the relationship between the stimulus intensity and the discomfort can be relatively evaluated for the subject as a personal characteristic.
[0400]
In determining the maximum applied current (maximum intensity) to the subject's muscle, the stimulation current is gradually increased by having the subject tensed the muscle, and the safety factor is the current value when the following conditions are observed. If the value multiplied by or exceeds the upper limit value (maximum current value or allowable upper limit value) of the stimulation current, the upper limit value is used.
[0401]
i) When a joint is subjugated by electrical stimulation
ii) When you feel painful.
[0402]
As for the relationship between the joint angle and the electrical stimulation, a predetermined stimulation output may be given to the target muscle and output setting or adjustment may be performed according to the joint angle at that time.
[0403]
In addition, if the resistance distribution calculated from the human body structure model database does not match the resistance value predicted from the body fat percentage of the subject, the insulation resistance value is corrected. Desirable for error reduction. For example, when the calculated resistance value is smaller than the body fat percentage of the subject, the proportional coefficient to be multiplied with each resistance value data is changed (the value of the proportional coefficient at the time of uncorrection is 1).
[0409]
【Example】
60 to 65 show an example of implementation of the apparatus according to the present invention, and this apparatus includes an I / O processing unit (mechanism unit) 100, an electrical stimulus generation unit 300, and an overall control unit 500. .
[0410]
Each component is as follows (indicated by a symbol in parentheses).
* I / O processing unit (100)
・ Electrode (101)
It is composed of a large number of electrode groups arranged and formed on a sheet-like substrate (see FIGS. 4 to 8), and an electrical signal sent from an electrical stimulation generator 300, which will be described later, through an electrode selector 104 is sent to each electrode. Supplied.
-Pressure and temperature detection unit (102) and element selection processing unit (103)
The sheet-like base material is composed of a large number of detection element parts (see FIGS. 9 to 12) arranged and formed on the upper layer of the electrode part 101. The element selection processing part 103 is used for selection of pressure and temperature detection elements. Is performed by matrix processing.
-Electrode selector (104)
As for the output system, a signal group (for example, “number of channels” × 2 signals) from the electrical stimulation generator 300 is sent to the output selector 104o, and is selected by receiving the control signal “Sel_o”. In contrast, an output signal is supplied through the buffer amplifier 104b. The buffer amplifier 104b is configured such that the output current is set by the control signal “Sc”. For the input system, signals from the electrodes are sent to the input selector 104i, and information selected in response to the control signal “Sel_i” is sent to the myoelectric detection unit 201 described later.
・ 3-axis position detector (105)
This is a detection unit for detecting the reference origin position for the subject, and is required for setting the coordinate system (three-dimensional orthogonal coordinate system, cylindrical coordinate system, axis coordinate system according to the joint structure, etc.) related to the subject. The As a detection method, an optical method, a magnetic method, a gyro sensor method, and the like are known.
* Myoelectric detection unit (201)
This is a part for recognizing myoelectric potential information acquired at the electrode designated by the electrode selection unit 104 by A / D conversion from an analog signal to a digital signal, and has the functions of the frequency analysis unit 1f and the FFT units 22a and 22b. is doing. The myoelectric detection may be performed at all times, or may be performed at a predetermined timing by a timer interruption or the like, or may be performed when a detection command is received. As for the data processing form after A / D conversion, only the number of rises per unit pulse when the threshold value specified for the pulse of the myoelectric waveform is exceeded may be used. There are various methods such as obtaining from integrated electromyogram (IEMG) obtained by time integration for conversion, or obtaining data by applying a spectrum compatible method after FFT processing of the electromyogram waveform.
* Shape recognition unit (202)
This is a part that receives pressure detection information obtained from the pressure and temperature detection unit 102 via the element selection processing unit 103 and recognizes the surface shape from the pressure distribution information. Note that the pressure detection information is all processed through matrix processing.
* Individuality recognition unit (203)
This is a part for recognizing the individuality of the subject based on information from the myoelectric detection unit 201 and the shape recognition unit 202, and the recognition result is sent to the overall control unit 500 described later. In the “recognition of individuality”, in addition to the above-described method, a method of discriminating from a component in a predetermined band (for example, 200 to 300 Hz) by frequency analysis of electromyogram information, or a myoelectric potential obtained for a plurality of kinematics A method of synthesizing information or shape recognition information and recognizing difference information between the configuration information and personal information as individuality (a method in which speech recognition technology is applied to myoelectric potential information) or the like can be used. It doesn't matter.
* Electrical stimulation generator (300)
The low frequency stimulus generator 301 and the interference wave stimulus generator 302 are provided according to the required number of channels, and these are placed under the command of the overall controller 500 described later.
* Evening position recognition unit (401)
This is a part for recognizing the position information of the even number based on information from the triaxial position detection unit 105, the myoelectric detection unit 201, and the shape recognition unit 202. That is, the information from the triaxial position detection unit 105 is used as the origin position information, and while the subject performs a specific operation, the shape recognition result at that time, the relationship with the kinematic exercise, the myoelectric potential information during exercise, The position of the even number is recognized based on the relationship such as the movement speed.
* Skeletal muscle placement recognition unit (402)
Based on the information from the kinematic position recognition unit 401, the three-dimensional human body shape data from the human body model processing unit 602 (the arrangement of bones and muscles in the kinematic pair is modeled with polygon data) is used. In this case, it is preferable to display the human body shape data on the monitor device 403 so that the human body shape data can be visually recognized.
* General control unit (500)
This is the part that controls the main functions of control, and includes the following elements.
[0411]
・ Stimulus selection function processing unit (501)
This is a part that sends a control signal for selecting the type of electrical stimulation to the electrical stimulation generator 300. That is, it defines whether to perform low-frequency electrical stimulation or interference wave electrical stimulation according to the depth from the skin for the muscle of the target site where electrical stimulation is performed. An electrical stimulus is generated, and for an interference wave, for example, an electrical stimulus is generated by combining four electrodes. In the electrical stimulation, the analysis result from the interference wave simulator (or a simplified version of the interference wave simulator) is used as described above.
[0412]
・ F. S. (Phantom Sensation) Stimulus Function Processing Unit (502)
This is a part that generates a signal pattern such as selection of an electrode and stimulation intensity used when applying a pseudo stimulus to the skin surface. In this case, an interference wave is used as described above. Further, in order to perform tactile sensation presentation according to command information from a force / tactile pattern recognition unit 505 described later, a control output is sent to a later-described use electrode change processing unit 507. In addition, when functional electrical stimulation is obstructed by phantom sensation stimulation, it is necessary to change the arrangement of the electrode pattern used and to quickly adopt another electrode pattern arrangement.
[0413]
・ Low-temperature burn prevention processing section (503)
It is provided to prevent low-temperature burns due to insulation contact resistance, and the pattern of the electrode used can be changed in a timely manner so that electrical stimulation is not applied using the same electrode for a certain period of time or the resistance of the skin surface Use of the electrode is prohibited when the value is greater than or equal to the threshold value or when the electrode is severely soiled. And the control command for that is sent out to the use electrode change process part 507 mentioned later. As described with reference to FIGS. 58 and 59, information from the myoelectric detection unit 201 is referred to (indicated by a broken line in the figure), and measures for preventing the occurrence of discomfort to the stimulus may be taken according to individual differences. preferable.
[0414]
・ Tracking function processing unit for change in muscle placement (504)
This is a part that performs control for changing the arrangement of the electrodes used in accordance with the movement of the muscle. For example, since the movement of the muscles during the pronation / extraction of the forearm is very large, it is necessary to change the use electrode arrangement in time in consideration of the movement of the muscles at that time. Therefore, the information on the muscle arrangement obtained by the skeletal muscle arrangement recognition unit 402, the driving amount of the kinematic pair predicted from the myoelectric detection unit 201 (refer to the operation prediction based on the predicted muscle tonus), and the actual kinematic position information The driving amount of the even number is recognized in real time based on the change in the number of electrodes, and the electrode position and the electrical stimulation waveform are changed according to the positional relationship between the pair.
[0415]
-Force / tactile pattern recognition unit (505)
This is the part that recognizes what kind of force / tactile sensation is given to the target part, and its control method is force / tactile feedback control (that is, it recognizes how much force / tactile sense is given to the command value. The control is such that the error obtained from the difference processing is zero). At that time, three-dimensional human body shape data from a human body model processing unit 602 described later is used. Note that the recognition result is sent to the phantom sensation stimulation function processing unit 502, the follow-up function processing unit 504 for muscle placement change, and the like.
[0416]
Selection / output setting processing unit (506)
Selection of electrode used (including selection of electrode position, use or non-use or prohibition), whether to apply electrical stimulation using the selected electrode, or to acquire myoelectric potential information using the electrode, Specify whether to take ground. In addition, output setting is performed when electrical stimulation is applied through the selection electrode. Note that the control signals Sel_o, Sel_i, and Sc are sent from the selection processing / output setting processing unit 506 via the selector unit 512 to various places, but the selector unit 512 is completely controlled by the selection processing / output setting processing unit 506. The condition is that the command is not received from a later-described used electrode change processing unit 507 (that is, the command is given priority for the selector unit 512).
[0417]
・ Use electrode change processing section (507)
This is a part for changing the electrode to be used according to various conditions, and its output is sent to the selector unit 512. Note that the output of this processing unit preferentially acts on the selector unit 512. In addition, the use electrode change command is sent from the phantom sensation function processing unit 502, the low-temperature burn prevention processing unit 503, the muscle placement change tracking function processing unit 504, the signal correction processing unit 508 described later, and the like. However, various electrode arrangement patterns are prepared according to the function of each processing unit.
[0418]
・ Signal correction processing unit (508)
The force-tactile feedback (state) expected based on the data obtained from the skeletal muscle arrangement recognition unit 402 is compared with the force-tactile feedback (state) actually occurring, and both match or the difference is reduced. As described above, the correction signal is sent to the stimulation pattern setting unit 510 and the used electrode change processing unit 507.
[0419]
・ Individual adaptation function processing unit (509)
This is a part having a function for changing the electrical stimulation waveform from its standard form based on the personality recognition data of the subject, and receives control information from the personality recognition unit 203 to the stimulation pattern setting unit 510. Is sent out.
[0420]
・ Stimulation pattern setting section (510)
Based on commands from the respective units (502 to 504, 508, 509) in the overall control unit 500, control signals related to the stimulation patterns (various stimulation signal patterns having different stimulation conditions described above) are sent to the electrical stimulation generation unit 300. Change the waveform.
[0421]
・ Simplified interference wave simulator (511)
The information obtained by the interference wave simulator 603 is used to reduce the frequency of use of the interference wave simulation by preliminarily creating a database of the positions of electrode candidates, the amount of even driving, and the like. The used electrode is determined by discriminating whether or not the even-number drive amount matches the even-number drive amount when the electrical stimulation is actually applied through the candidate electrode.
* I / O processing section
It includes all devices necessary for setting force / tactile patterns (command patterns) and displaying 3D polygon data as images. For example, in addition to an output device such as a computer input device 601 (keyboard, mouse, etc.), a monitor device 403, a head-mounted display device, a printer, etc., means necessary for data input / output processing related to the database of the human body structure (model) described above Is also included.
* Human body model processing unit (602)
This corresponds to the model construction / calculation processing unit 1Ae. In constructing the model, as described above, it is possible to obtain the data of the skeleton and muscle arrangement of the target site from the tomographic plane information obtained by helical CT, MRI or the like.
* Interference wave simulator (603)
As described above, it is used for selection of electrode candidates for use when applying interference wave electrical stimulation.
[0422]
Note that the application to the force / tactile sensation presentation apparatus requires processing relating to the generation of an object in the virtual reality space and the command of force / tactile sensation presentation. For example, in application to a game machine, as shown in FIGS. In addition to the above components, the following elements are required in the game machine 700 (the numbers in parentheses indicate symbols).
Stereo image output unit (701)
This is a part that outputs a stereo video (signal) according to the story development, and the video signal is sent to a visual display device 702 such as a head-mounted display (or 3D scope) and provided to the subject. Note that sound signals such as sound effects are also provided to the subject using sound output means such as headphones or speakers incorporated in the visual display device, but illustration is omitted. Data necessary for the progress of the game is provided by an information recording medium (not shown).
[0423]
-Forced tactile generation command part (703)
This is a part for giving a command about generation of a force / tactile sensation, and a force / tactile sensation presentation command is issued in a scene where the force / tactile sensation is required in the story development, and this is sent to the force / tactile pattern recognition unit 505. Thereby, the force / tactile pattern recognition unit 505 interprets the contents of the presentation command. In addition, recognition of the even position and acquisition of position information are performed for the target part of the force / tactile sensation presented by a command sent from the command section to the even position recognition section 401.
[0424]
Next, a control method of the force / tactile sensation presentation apparatus using functional electrical stimulation will be described with reference to the flowcharts shown in FIGS.
[0425]
FIG. 66 shows the outline of the flow of the entire control, and that a database in which data of kinematic actions and electromyograms are linked to a database related to the human body structure of the subject has been constructed, and It is a premise that the result of the interference wave simulation using the database is also made into a database. As described above, the skeleton data and muscle arrangement data corresponding to the joint angle are acquired by helical CT, and the data of muscle arrangement is acquired by helical MRI or the like and stored in a database.
[0426]
In starting the process, first, an effector including the electrode unit 101, the pressure and temperature detection unit 102, and a head mounted display are attached to the subject. At that time, the entire skin needs to be firmly covered with the effector for the target site of electrical stimulation.
[0427]
In step S1, a calibration process (details thereof will be described with reference to FIG. 67) necessary for obtaining the reference surface pressure information is performed.
[0428]
Then, in the next step S2, a calibration process necessary for acquiring information about the change in surface pressure due to the joint angle is performed (this is necessary for recognition regarding centripetal contraction and centrifugal contraction, and details thereof will be described with reference to FIG. 68). To do.)
[0429]
In the next step S3, after a calibration process for the muscle work (details will be described with reference to FIGS. 69 and 70), in the next step S4, the data obtained in the previous steps S1 to S3 and the above-described data are obtained. Data generated according to the human body structure model algorithm, and polygon data generated using information such as helical CT and MRI (required for recognizing how the muscle centerline and shape change due to joint movement) For details, refer to FIG. 45 to FIG. 47 and the related description.) In this case, the reference of the link is the joint angle, and the joint angle estimated from the shape of the pair is also taken into consideration.
[0430]
In step S5, the force / tactile sense presenting instruction is interpreted. In other words, the command is sent from the force / tactile sensation generation command unit 703 to the force / tactile pattern recognition unit 505 to interpret the contents (the upper command is reduced to a lower-dimensional command as it goes to the lower layer components To go.).
[0431]
For example, the setting conditions of force / tactile pattern include exercise pattern, speed, work amount, etc. Based on this, the stimulation position, strength, type of skin surface (for example, receiving a sword wound, landing, etc. or bounced) Various parameters are defined, such as a type according to the difference in touch. In setting the stimulation position, it is necessary to specify according to the story development in the case of a game machine, and data for that is stored in an information recording medium in advance.
[0432]
In the next step S6, the control is performed in consideration of the relationship between the even position and gravity and the load. That is, the position of the even number is recognized by a command sent from the command / generation unit 703 for generating force / tactile sensation to the position detecting unit 401, but the influence of gravity on the subject becomes a problem at that time.
[0433]
There are the following three control methods (here, when one joint is driven by a plurality of muscles, for example, the upper arm is assumed).
[0434]
-A method for constructing a database on joint contraction and muscle contraction that assists in driving the joint in advance, and using the database for electrical stimulation, and always controlling the paired muscles and controlling them.
・ A database on the relationship between the positions of joints and even pairs and gravity is built in advance, and when performing electrical stimulation, the database is used to identify the muscles that make a pair according to the joint position with respect to gravity, and tension them. How to control
・ For a load (for example, a heavy object held by a hand), a driving muscle to be stimulated is added to the database, and the electric muscle is selected and tensioned. How to control.
[0435]
In the next step S7, the electrodes to be used and the stimulation conditions are set according to the force / tactile sensation presentation command, and electrical stimulation is applied (the details will be described with reference to FIG. 71).
[0436]
FIG. 67 is a flowchart relating to the acquisition (calibration) of the reference plane pressure information. First, in step S1, information on the kinematic position that is the basis based on information from the triaxial position detection unit 105 using a magnetic sensor or the like. In the next step S2, information on the visual position (viewpoint) of the subject is acquired in the video on the visual display device (head mounted display or the like) using the information from the triaxial position detection unit 105.
[0437]
In the next step S3, based on the kinematic position obtained in the previous step S1, three-dimensional polygon data about the human body part from the kinematic pair to its end is generated according to the algorithm of the human body structure model, and then the process proceeds to the next step S4. Thus, an image corresponding to the polygon data is displayed on the visual display device 702.
[0438]
In step S5, the target person imitates the operation state so that his / her even number (image) overlaps the polygon data image. At that time, voice navigation is also performed at the same time.
[0439]
In the next step S6, it is determined whether or not the imitation of the operation has been completed based on the degree of overlap of the images. The process proceeds to S8, but if the imitation is not completed, the process returns to Step S5.
[0440]
In step S8, the numerical data acquired in the previous step S7 is stored and registered as a reference value (for example, “100” when percentage display is adopted), and a database of reference surface pressure information is created based on this.
[0441]
FIG. 68 is a flowchart regarding the acquisition (calibration) of surface pressure information according to the joint angle, and steps S1 and S2 are the same as steps S1 and S2 in FIG.
[0442]
In step S3, based on the kinematic position and joint angle obtained in the previous step S1, three-dimensional polygon data for the human body part from the kinematic pair to its end is generated according to the algorithm of the human body structure model, and then in the next step S4. Then, the video corresponding to the polygon data is displayed on the visual display device 702.
[0443]
In step S5, the target person imitates the operation so that his / her even number (image) overlaps the polygon data image (in this case, navigation by voice is also performed at the same time). In the next step S6, it is determined whether or not the imitation of the operation is completed based on the overlapping state of the images. When the operation is completed, the process proceeds to the next step S7, and the pressure distribution at that time is acquired from the information of the pressure detection element group of the effector Thereafter, the process proceeds to step S8, but if the imitation is not completed, the process returns to step S5.
[0444]
In step S8, a location and a pressure value having a large change are acquired from the pressure information obtained in the previous step S7 with reference to the data in the reference surface pressure information database created in step S8 of FIG. For example, when comparing with the reference surface pressure information, only a predetermined place is selected from the upper places where the pressure detection value is rich, and those positions and pressure values are stored and registered. Create a database of surface pressure information.
[0445]
In step S9, it is determined whether or not the processing in steps S3 to S8 has been completed for all the angles of the joint. If not completed, the process returns to step S3 to change the joint angle from the previous value and then step. The processes from S3 to S8 are repeated.
[0446]
Thus, the change in the surface pressure when the joint angle is changed as a parameter can be acquired as a relative value with respect to the reference surface pressure.
[0447]
69 and 70 are flowcharts relating to the acquisition (calibration) of myoelectric (pattern) information based on the work of the muscle, and steps S1 to S6 in FIG. 69 are the same as the respective steps in FIG.
[0448]
In step S7 of FIG. 70, after setting so that myoelectric potential information can be acquired through the electrode part 101 of the effector, the pattern of the myoelectric waveform is acquired a plurality of times.
[0449]
In step S8, it is determined whether or not the information obtained in the previous step has already been stored and registered. If so, the process proceeds to step S9. If not registered, the process proceeds to step S10.
[0450]
In step S10, the data obtained in the previous step is averaged and stored and stored as a reference value (for example, “100” when percentage display is adopted). Create a database related to myoelectric pattern information. Then, the process proceeds to the next step S11, and after wearing a device for applying a predetermined load to the target part, the process returns to step S3.
[0451]
In step S9, data having a large change is acquired from the myoelectric potential information obtained in the previous step S7 based on the no-load myoelectric pattern information created in step S10. For example, when comparing with a myoelectric pattern at no load, select a predetermined number of data with significant changes in potential from the top, store and register them, and based on this, relate to myoelectric pattern information at load Create a database.
[0452]
In this way, the myoelectric potential information at the time of loading (applying) can be acquired as a relative value with respect to the myoelectric potential information at the time of no load of the target part.
[0453]
FIG. 71 is a flowchart showing an example of a procedure for presenting a force sensation by the generation of an electrical stimulus.
[0454]
First, the joint drive angle is set in step S1, and then the even shape is grasped based on the pressure detection in the next step S2. In step S3, three-dimensional polygon data of the target part (or shape data of the internal structure based on the tomographic analysis result) is generated. Then, in the next step S4, the stimulation region (in the explanation of the interference wave simulation described above) for the muscle to be driven is generated. Set points, planes, and solids) to apply electrical stimulation.
[0455]
As a result, the even number is actually driven in step S5. In the next step S6, it is determined whether or not the driving angle set in the previous step S1 has been achieved. If is not achieved, the process returns to step S2.
[0456]
In addition, since many muscles cooperate with each other in the upper arm and the hand in the movement of the human body, it is very difficult to individually apply the interference wave electrical stimulation to all the muscles. Therefore, it is desirable to use both low-frequency electrical stimulation and interference wave electrical stimulation (including a stimulation method using phantom sensation for tactile presentation) according to the arrangement of the muscles that operate the kinematic pair.
[0457]
FIG. 72 is a flowchart showing an example of the acquisition of the myoelectric potential information and the storage / learning process for the muscle motion prediction. In the above-described apparatus configuration, the parts directly related to each other are the input selector (104i), the myoelectric detection unit (201), the output selector (104o), and the electrical stimulation generator (300).
[0458]
First, step S1 is a branch process for the preparation process, and the process proceeds to an unfinished step among steps S2 to S4.
[0459]
In step S2, the target person wears an effector (for example, a suit-shaped one) and takes an anatomical standing posture, and the effector is set so as not to wrinkle the surface that contacts the skin surface. After that, the surface shape data of the subject is acquired according to the shape recognition method using (I) pressure detection. The reason why this data acquisition process is necessary is that it is impossible to dynamically grasp the shape change associated with the bulging of the muscle or the like only with steps S3 and S4 described later.
[0460]
Further, in step S3, as described in (II) Construction of database using information on bone cross section, muscle and fat cross section, data acquisition and database creation of tomographic cross section using helical CT, MRI, optical CT, etc. Process.
[0461]
In step S4, a database relating to the human body structure of the subject is constructed (depending on the method for constructing the numerical model of the mechanical structure described above).
[0462]
In step S5, it is determined whether or not all the preparation processes have been completed. When finished, the process proceeds to the next step S6, but when not completed, the process returns to step S1.
[0463]
In step S6, after grasping the physique of the subject from the data obtained in the previous steps S2 to S4, the muscle arrangement is recognized in the next step S7. That is, since the subject's body shape is obtained in step S4, the approximate muscle arrangement can be understood by this, but the subject's external shape obtained in step S2 and the internal structure data obtained in step S3 can be used for more accuracy. It is possible to achieve a high-accuracy muscle recognition and model accuracy.
[0464]
In step S8, after using the method by the said interference wave simulator, the use electrode in an effector is selected, and in next step S9, a subject moves a muscle by own will. As an operation pattern in that case, the following operation examples are given in the case of the forearm, for example.
[0465]
・ Elbow extension, pronation, 2nd to 5th finger DIP, PIP, MP flexion, 1st finger IP, MP flexion with hand in front and fist
・ Hand gripping the iron array and lifting it, elbow flexion, palm flexion of the wrist during finger flexion
・ Confrontation (flat state), MP abduction of the 2nd to 5th fingers, CM buckling abduction of the 1st finger in a state where the palm surface is flattened by putting the hand forward
・ Palm flexion of wrist during finger extension from bent wrist
-Wrist dorsiflexion from a state where the wrist is bent upward
・ DIP, PIP, MP extension and MP adduction of 2nd to 5th fingers in hand
-Confrontation (curved state) with the thumb at the base of the little finger, CM first flexion adduction of the first finger.
[0466]
In step S10, the myoelectric detection unit 201 obtains myoelectric potential information and performs frequency analysis after selecting the electrodes to be used under the control of the input selector 104i. The data acquired at this time (this is defined as “master EMG”) includes the frequency components shown below.
[0467]
・ Ingredients related to muscle tonus
・ Total component related to motor units.
[0468]
For the master EMG, data indicating the operation pattern at that time is linked and stored in a predetermined storage means.
[0469]
In step S11, the target muscle is operated only by electrical stimulation, and in step S12, myoelectric potential information at that time is acquired and frequency analysis is performed. That is, the data acquired in this case (this is defined as “slave EMG”) includes the frequency components shown below.
[0470]
・ Electric stimulation signal components
・ Total component related to motor units.
[0471]
Therefore, by subtracting the component of the electrical stimulation signal from the slave EMG (this is known as the original signal of the electrical stimulation and can be directly calculated by FFT analysis of the signal), the total component of the motor unit is obtained. If removed from the master EMG, only muscle tonus components can be extracted. If this is stored and learned in association with the motion pattern, muscle motion prediction can be easily performed from the subsequent EMG data. That is, the motion can be predicted from the myoelectric potential information acquired from the electrodes without considering the characteristics of the motor unit.
[0472]
For example, when EMG measurement is performed before and after performing push-ups 100 times, the acquired EMG data is different because the effect of muscle fatigue has occurred on the exercise unit. Therefore, it is extremely difficult to predict the motion directly from the EMG data as in the conventional method. However, according to the present invention, the muscle tonus component obtained by removing the sum component related to the motor unit from the EMG data. Based on this, it becomes possible to predict the operation.
[0473]
Therefore, according to the above-described apparatus, the following advantages can be obtained.
[0474]
-Since an exoskeleton type mechanism using a conventional shaft arm or the like is not required, it is advantageous for weight reduction, and the subject can use it as if wearing clothes.
[0475]
・ Low power consumption for generating external force. In other words, a power source such as a servo motor or a pneumatic pressure source of an air cylinder is not required, and a joint can be driven or a tactile sensation can be provided by applying a stimulus to the subject's muscle.
[0476]
Since there is no noise during driving and a force-tactile sensation can be presented without sound, for example, there is no adverse effect on sound effects and environmental sounds when used for game machines.
[0477]
・ By using phantom sensation, it is possible to express a pseudo stimulus to the skin surface.
[0478]
-Muscle motion prediction can be performed based on muscle tonus prediction data without being affected by the characteristics and state of the motor unit.
[0479]
【The invention's effect】
As apparent from the above description, according to the invention according to claim 1, while applying a certain electrical stimulus to the muscle of the subject, the myoelectric potential information of the muscle is acquired and the frequency analysis is performed. Before the muscle actually operates, the basic data resulting only from the contraction of myocytes is obtained in advance, and then the data obtained by removing the basic data from the data obtained by frequency analysis of the myoelectric potential information is calculated. Muscle motion can be predicted. Accordingly, since motion prediction can be performed based on data relating to muscle tonus obtained from myoelectric potential information, the prediction accuracy is hardly affected by various factors including the degree of muscle fatigue.
[0480]
According to the second aspect of the present invention, the frequency component of the motion set state signal of the multi-muscle fiber by ATP is removed from the frequency component of the signal detected from the electrode when the subject tries to move the muscle on his / her will. Based on the obtained data, muscle motion prediction can be performed.
[0481]
According to the third aspect of the present invention, muscle tonus prediction data can be obtained without considering the characteristics of the motor unit in the myoelectric potential information.
[0482]
According to the invention which concerns on Claim 4, recognition of individuality is attained by acquiring the data concerning a subject's individual muscle characteristic.
[0483]
According to the fifth aspect of the present invention, it is possible to quickly predict a motion pattern by storing and learning muscle tonus prediction data in association with the motion pattern of the subject in advance.
[0484]
According to the inventions according to claims 6, 11 and 16, since the motion of the muscle can be predicted before the electrical stimulation is actually applied when the electrode is selected, the motion that does not follow the intention of the subject by the electrical stimulation. It is possible to prevent the occurrence of unsatisfactory stimulation or change of movement in the muscle movement.
[0485]
According to the invention which concerns on Claim 7, 12, the simplification of a structure can be achieved by laminating | stacking and forming an electrode part and a shape information acquisition part on the same sheet-like base material.
[0486]
According to the invention which concerns on Claim 8, 13, an electrode group can be efficiently arrange | positioned on a sheet-like base material.
[0487]
According to the invention which concerns on Claim 9, 14, an electrode group can be arrange | positioned without waste with respect to the skin surface which is not caught by operation | movement of a joint.
[0488]
According to the invention which concerns on Claim 10, 15, while mounting an effector to a human body becomes easy, the stuffiness at the time of mounting | wearing can be prevented.
[0489]
According to the invention which concerns on Claim 17, since the effector suitable for the shape can be partially created for every even number, it is easy to wear.
[0490]
According to the eighteenth aspect of the invention, it is possible to grasp the outer shape of the subject based on the pressure detection information by the effector worn without requiring imaging means such as a camera.
[0491]
According to the nineteenth aspect of the invention, it is possible to improve the model accuracy by obtaining a database including internal structures such as bones and muscles.
[0492]
According to the twentieth aspect of the present invention, it is possible to easily identify the streak by image processing based on the difference in color difference, and to easily distinguish the streak.
[0493]
According to the twenty-first aspect of the present invention, three-dimensional shape data can be created by easily acquiring point information that defines the shape of a bone or muscle.
[0494]
According to the twenty-second aspect of the present invention, the dynamic change of the three-dimensional shape data can be easily expressed.
[0495]
According to the invention of claim 23, the muscle characteristics of the subject can be grasped from the quantitative ratio of the red and white stripes.
[0496]
According to the inventions according to claims 24 to 26, it is possible to grasp the electrode position suitable for the stimulation region of the target part by performing the simulation of the interference wave in advance when selecting the electrode to be used, and thereby the actual electrical stimulation. An ideal interference wave stimulus can be obtained at the time of application.
[0497]
According to the twenty-seventh aspect of the present invention, the output of electrical stimulation can be set according to the arrangement of the electrodes selected from the electrode candidates, and the occurrence of low-temperature burns or the like can be prevented.
[0498]
According to the invention of claim 28, by storing and registering the electrode candidates in advance, it is possible to immediately select the electrode to be used from the electrode candidates when actual electrical stimulation is applied.
[0499]
According to the invention of claim 29, it is possible to improve the accuracy of muscle tonus prediction by detecting myoelectric potential information from the same position as the electrode position finally selected for application to electrical stimulation and performing frequency analysis. .
[0500]
According to the invention of claim 30, it is possible to accurately predict the motion of the subject's joint.
[0501]
According to the invention of claim 31, it is possible to control so that the application of electrical stimulation using the same electrode and the acquisition of myoelectric potential information do not compete simultaneously.
[Brief description of the drawings]
FIG. 1 is a diagram showing a basic configuration of an apparatus according to the present invention.
FIG. 2 is a diagram illustrating a configuration example of a force / tactile sensation presentation apparatus using functional electrical stimulation.
FIG. 3 is a view for explaining a configuration of an electrode sheet used in the effector together with FIGS. 4 to 14, and this figure is a view showing a state in which the electrode sheet is wound around an arm.
FIG. 4 is an exploded perspective view of an electrode sheet.
FIG. 5 is a diagram partially showing the sheet 4 on the most skin side constituting the electrode layer 3A.
6 is a view partially showing a conductive sheet 5 constituting an electrode layer 3A. FIG.
7 is a view partially showing a wiring sheet 6 constituting the electrode layer 3A. FIG.
FIG. 8 is a view partially showing the top sheet 7 constituting the electrode layer 3A.
FIG. 9 is a diagram partially showing a row electrode sheet 8 constituting the pressure-sensitive / temperature-sensitive layer 3B.
FIG. 10 is a diagram partially showing a pressure-sensitive / temperature-sensitive sheet 9 constituting the pressure-sensitive / temperature-sensitive layer 3B.
FIG. 11 is a diagram partially showing a column electrode sheet 10 constituting a pressure-sensitive / temperature-sensitive layer 3B.
12 is a view partially showing an insulating film 11 constituting the pressure-sensitive / temperature-sensitive layer 3B. FIG.
FIG. 13 is an enlarged cross-sectional view showing a part of an electrode sheet.
FIG. 14 is a schematic cross-sectional view showing an electrode sheet wound around an arm.
15 is a view for explaining detection of a bent state of a finger in the data glove together with FIG. 16, and this figure shows a back surface of the finger part of the data glove. FIG.
FIG. 16 is a diagram showing a state in which a finger is bent.
FIG. 17 is a diagram schematically illustrating a main configuration of a data glove including a resistance pattern, a pressure and a temperature detection unit.
FIG. 18 is an explanatory diagram of body type classification of a human body.
FIG. 19 is a diagram illustrating definitions of various quantities in a reference body model.
FIG. 20 is an explanatory diagram of a two-dimensional coordinate plane constituting a body shape coordinate space.
FIG. 21 is a diagram illustrating a state in which a variable of the length ratio is set on the Z axis in the body shape coordinate space.
FIG. 22 is a diagram illustrating a state in which variables of length ratio and weight ratio are set on the Z axis in the body shape coordinate space.
FIG. 23 is an explanatory diagram of image data acquired from a subject and its expansion / contraction operation.
FIG. 24 is an explanatory diagram regarding slice processing on the three-dimensional data of the subject together with FIG. 25, and shows the subject (image) in the standing posture and the slice interval.
FIG. 25 is a diagram illustrating a relationship between slice processing and a tomographic plane for three-dimensional data of a subject.
FIG. 26 is a diagram showing an example of arrangement of points in a polar coordinate system (r, θ) set on an XY plane constituting a body shape seat space;
FIG. 27 is an explanatory diagram of a process of specifying a subject's body shape by calculating the center of gravity of a polygon.
FIG. 28 is a diagram illustrating a state in which a function value corresponding to the center of gravity G (Xg, Yg) is obtained.
29 is a flowchart showing an example of processing related to input of basic data related to the body and data processing together with FIG. 30, and this figure shows the first half of the processing.
FIG. 30 shows the latter half of the process.
FIG. 31 is a flowchart showing a processing example related to acquisition of three-dimensional data about a human body, generation of a wire frame model related to exercise, and acquisition of barycentric position data associated with exercise.
FIG. 32 is a schematic explanatory diagram of slice processing relating to three-dimensional data of a subject.
FIG. 33 is a diagram illustrating a dependency relationship between a human body polygon DB and its configuration DB.
FIG. 34 is a flowchart showing a processing example related to generation of a human body polygon DB.
FIG. 35 is a diagram for explaining initial learning together with FIGS. 36 to 40, and shows a pressure detection sheet wound in a cylindrical shape.
FIG. 36 is a diagram showing another example of a pressure detection sheet wound in a cylindrical shape.
FIG. 37 is a diagram for describing element selection for a pressure detection sheet;
FIG. 38 is a diagram showing a set of reference cylinders that are used by being inserted into a pressure detection sheet.
FIG. 39 is a flowchart showing an example procedure of initial learning.
FIG. 40 is a flowchart illustrating an example of a procedure for acquiring a shape from pressure distribution information using a pressure detection sheet.
FIG. 41 is a flowchart for explaining the shape recognition of the upper limbs together with FIGS. 42 to 44, and is a flowchart showing an example of the procedure for shape recognition.
FIG. 42 is an explanatory diagram showing a state in which a visual display device is mounted on a subject and the movement of the upper limb to be imitated is projected on the device.
FIG. 43 is a diagram schematically showing an upper limb of a human body.
FIG. 44 is a graph showing several examples of pressure patterns.
FIG. 45 is a flowchart for explaining a processing example for creating a shape model including the internal structure of the target part from the tomographic plane data together with FIGS. 46 and 47, and this figure shows a start part of the processing.
FIG. 46 shows an intermediate part of processing.
FIG. 47 shows an end part of processing.
FIG. 48 is an explanatory diagram showing assignment of the number of polygon points to the contour line length for each tomographic plane and a graphic (crescent shape) on that plane.
FIG. 49 is an explanatory diagram showing a state in which a polygon point is generated on a contour line by a crescent moon shape, its center of gravity G, and an equiangular dividing line centered on the center of gravity G.
FIG. 50 is an explanatory diagram of a calculation process of a surface centroid position.
FIG. 51 is an explanatory diagram showing a polygon point obtained as an intersection of an equiangular dividing line centered on the surface center of gravity and a contour line;
FIG. 52 is a waveform diagram for explaining the definitions of various quantities for stimulation conditions.
FIG. 53 is a diagram for explaining the interference wave simulation together with FIG. 54. FIG. 53 is a schematic diagram showing the relationship between one point on the skin surface S and an internal target site XP.
FIG. 54 is a schematic diagram showing a relationship between three points on the skin surface S and an internal target site XP.
FIG. 55 is a diagram for explaining a method for acquiring a multi-muscle fiber motion set signal by ATP together with FIG. 56, and is a diagram schematically showing an example of the apparatus.
FIG. 56 is a graph conceptually showing frequency characteristics after FFT.
FIG. 57 is a diagram conceptually illustrating a configuration example in a case where electrical stimulation through an electrode and acquisition processing of myoelectric potential information from the electrode are performed in a time-sharing process.
FIG. 58 is a flowchart showing an example of a subjective evaluation method using fat and stimulation of the human body together with FIG. 59, and this figure shows the first half thereof.
FIG. 59 is a diagram showing the latter half of the process.
FIG. 60 shows an embodiment of the present invention together with FIGS. 61 to 72, and is a block diagram showing the overall configuration of the apparatus.
FIG. 61 is a diagram mainly illustrating an I / O processing unit.
FIG. 62 is a diagram illustrating a configuration of an overall control unit.
FIG. 63 is a diagram illustrating a relationship between the overall control unit and its peripheral units.
FIG. 64 shows an application example to a game machine as a force / tactile sensation presentation apparatus together with FIG. 65, and this figure is a block diagram showing an overall configuration of the apparatus.
FIG. 65 is a diagram showing only the configuration of the main part.
66 is a flowchart for explaining the control method of the apparatus together with FIGS. 67 to 71, and this figure is a flowchart showing an overall flow of control. FIG.
FIG. 67 is a flowchart relating to acquisition of reference plane pressure information;
FIG. 68 is a flowchart relating to the acquisition of surface pressure information according to the joint angle.
FIG. 69 is a flowchart related to the acquisition of myoelectric information based on the amount of work of a muscle, together with FIG.
FIG. 70 shows the latter half of the process.
FIG. 71 is a flowchart showing an example of a procedure for presenting a force sensation due to generation of an electrical stimulus.
FIG. 72 is a flowchart showing an example of acquiring myoelectric potential information and storing / learning processing;
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Motion capture apparatus, 1a ... Effector, 1as ... Shape information acquisition part, 1ad ... Electrode part, 1b ... Electrical stimulus generation part, 1c ... Shape recognition part, 1d ... Control means, 1f ... Frequency analysis part, 1A ... Power Tactile sense presentation device, 1Ad ... effect device, 2 ... electrode

Claims (31)

In the motion capture device using the myoelectric potential information for detecting the myoelectric potential information of the subject and obtaining the data indicating the movement change of the muscle,
(A) having an electrical stimulus generator for supplying electrical signals to a plurality of electrodes used in contact with the skin surface;
(B) Of the electrodes in (a) above, by calculating difference data between the frequency component related to the output signal of electrical stimulation to the anode side electrode and the frequency component of the signal detected from the cathode side electrode A frequency analysis unit that analyzes the frequency component of the signal generated from the muscle by electrical stimulation;
(C) The difference data related to the frequency component obtained by the frequency analysis unit is stored as the data of the frequency component related to the motion set state signal of the multi-muscle fiber by ATP, and then the data is transferred from the electrode. Comprising a control means for calculating a result of subtraction from the frequency component related to the detected myoelectric potential information and predicting muscle movement;
A motion capture device that uses myoelectric potential information. In the motion capture device using the myoelectric potential information according to claim 1,
The muscle actually operates by the control means predicting muscle tonus based on information obtained by removing the frequency component of the motion aggregate state signal of multi-muscle fibers from the frequency component related to the myoelectric potential information detected from the electrode. A motion capture device using myoelectric potential information characterized by predicting the motion before performing. In the control method of the motion capture device using the myoelectric potential information according to claim 2,
After detecting the first myoelectric potential information obtained when the subject operates the muscle according to his / her intention and the second myoelectric potential information obtained when operating the muscle by electrical stimulation,
By subtracting the component obtained by removing the frequency component of the electrical stimulation signal from the frequency component related to the second myoelectric potential information from the frequency component related to the first myoelectric potential information, the frequency component data related to myogenic tonus is obtained. A method for controlling a motion capture device using myoelectric potential information, characterized in that this is used as data for predicting muscle tonus. In the motion capture device using the myoelectric potential information according to claim 2,
A motion capture device using myoelectric potential information, characterized in that the control means obtains data relating to the muscle characteristics of the subject individual from the difference data obtained by the frequency analysis unit. In the control method of the motion capture device using the myoelectric potential information according to claim 3,
After storing and learning the muscle tonus prediction data calculated from the first myoelectric potential information and the second myoelectric potential information in association with the motion pattern of the subject,
Compare and collate the muscle tonus component data calculated from the frequency analysis result of the myoelectric potential information when the subject tries to move the muscles with the prediction data that has already been memorized and learned. A motion capture device control method using myoelectric potential information, wherein a motion pattern of a muscle corresponding to the data is predicted. In an electrical stimulation device including an effector for applying electrical stimulation to muscles,
(B) The effector includes an electrode unit composed of an electrode group used in contact with the skin surface, and a shape information acquisition unit that acquires information indicating the shape and shape change of the skin surface.
(B) having an electrical stimulation generator for supplying an electrical signal of low frequency electrical stimulation or interference wave electrical stimulation to the electrode group of (a) above;
(C) Calculate the difference data between the frequency component of the electrical stimulation signal to the anode side electrode and the frequency component of the signal detected from the cathode side electrode among the electrodes constituting the electrode group of (a) above. A frequency analysis unit that analyzes the frequency component of the signal generated from the muscle by electrical stimulation,
(D) having a shape recognition unit for recognizing the shape of the skin surface from the information obtained by the shape information acquisition unit;
(E) Constructing a database showing the muscle arrangement of the human body related to the subject, and electrical stimulation of the electrode group based on the information on the muscle arrangement and the recognition result of the muscle shape and shape change from the shape recognition unit A control means for selecting an electrode to which a signal is to be supplied;
(F) The control means stores the difference data related to the frequency component obtained by the frequency analysis unit as data of the frequency component related to the motion set state signal of the multi-muscle fiber by ATP, and thereafter the data Comparing the myoelectric potential information detected from the electrode and the electrostimulation information before the electric signal is supplied from the electrical stimulation generator to the selected electrode,
An electrical stimulator characterized by. The electrical stimulation device according to claim 6,
An electrical stimulation device, wherein an electrode part and a shape information acquisition part are formed with a multilayer structure on a sheet-like base material. The electrical stimulation device according to claim 7,
The electrode group constituting the electrode unit is disposed only in the region on the sheet-like base material corresponding to the muscle start point and muscle abdominal area, or the vicinity thereof, or the range extending from the muscle start point to the muscle end point. An electric stimulator characterized by being made. The electrical stimulation device according to claim 7,
An electrical stimulation device, characterized in that an electrode group constituting an electrode portion is not disposed in a region of the skin surface that is involved by movement of a joint. The electrical stimulation device according to claim 7,
The effector has a shape used by being worn on the human body, and a material having air permeability and stretchability is used in a place where the electrode group constituting the electrode portion is not arranged in the sheet-like base material. An electrical stimulation device characterized by being provided. In a force / tactile sensation presentation device using electrical stimulation, equipped with an effector for presenting force sense and tactile sensation by applying electrical stimulation to muscles,
(B) The effector includes an electrode unit composed of an electrode group used in contact with the skin surface, and a shape information acquisition unit that acquires information indicating the shape and shape change of the skin surface.
(B) having an electrical stimulation generator for supplying an electrical signal of low frequency electrical stimulation or interference wave electrical stimulation to the electrode group of (a) above;
(C) Calculate the difference data between the frequency component of the electrical stimulation signal to the anode side electrode and the frequency component of the signal detected from the cathode side electrode among the electrodes constituting the electrode group of (a) above. A frequency analysis unit that analyzes the frequency component of the signal generated from the muscle by electrical stimulation,
(D) having a shape recognition unit for recognizing the shape of the skin surface from the information obtained by the shape information acquisition unit;
(E) Constructing a database showing the muscle arrangement of the human body related to the subject, and electrical stimulation of the electrode group based on the information on the muscle arrangement and the recognition result of the muscle shape and shape change from the shape recognition unit A control means for selecting an electrode to which a signal is to be supplied;
(F) The control means stores the difference data related to the frequency component obtained by the frequency analysis unit as data of the frequency component related to the motion set state signal of the multi-muscle fiber by ATP, and thereafter the data By comparing the myoelectric potential information detected from the electrode and the electrical stimulation signal to the target electrode by supplying an electrical signal to the selected electrode from the electrical stimulation generating unit,
A force-tactile sensation presentation apparatus using electrical stimulation characterized by the above. The force / tactile sensation presentation apparatus using electrical stimulation according to claim 11,
A force-tactile sensation presentation apparatus using electrical stimulation, wherein an electrode part and a shape information acquisition part are formed on a sheet-like base material with a multilayer structure. The force-tactile sensation presentation apparatus using electrical stimulation according to claim 12,
The electrode group constituting the electrode unit is arranged only in the region on the sheet-like base material corresponding to the muscle start point and muscle abdomen of the electric stimulation target or the vicinity thereof or the range extending from the muscle start point to the muscle end point. A force-tactile sensation presentation apparatus using electrical stimulation, characterized in that The force-tactile sensation presentation apparatus using electrical stimulation according to claim 12,
A force-tactile sensation presentation apparatus using electrical stimulation, characterized in that an electrode group constituting an electrode portion is not arranged in an area of a skin surface that is involved by an operation of a joint. The force-tactile sensation presentation apparatus using electrical stimulation according to claim 12,
The effector has a shape used by being worn on the human body, and a material having air permeability and stretchability is used in a place where the electrode group constituting the electrode portion is not arranged in the sheet-like base material. A force-tactile sensation presentation apparatus using electrical stimulation, characterized in that In the control method of the electrical stimulation device used for giving electrical stimulation to the muscle or presenting force sense or tactile sense by the electrical stimulation,
(B) A database showing the muscle arrangement of the human body related to the subject is built in advance,
(B) By setting the effector on the subject, the electrode group constituting the effector is set to be in contact with the skin surface,
(C) Of the electrodes constituting the electrode group of (b) above, difference data between the frequency component related to the output signal of the electrical stimulation to the anode side electrode and the frequency component of the signal detected from the cathode side electrode By calculating the frequency component of the signal generated from the muscle by electrical stimulation, and storing the difference data as the data of the frequency component related to the motion set state signal of the multi-muscle fiber by ATP,
(D) Acquire information indicating the shape and shape change of the skin surface to perform shape recognition,
(E) After selecting the electrode to which the electrical stimulation signal should be supplied from the electrode group based on the information on the muscle arrangement of the subject obtained from the database and the shape recognition result,
(F) By comparing the myoelectric potential information detected from the electrode selected in (e) with the data stored in (c), the movement is predicted before the movement of the muscle. A control method for an electrical stimulation apparatus, comprising supplying an electrical stimulation signal to the electrical stimulation apparatus. The method of controlling an electrical stimulation device according to claim 16,
A control method for an electrical stimulation device, wherein an electrical stimulator is applied by attaching an effector having an electrode portion only to a portion corresponding to a pair of joints. The method of controlling an electrical stimulation device according to claim 16,
After mounting an effector having a multilayer structure on an electrode part and a pressure detection part on a sheet-like base material, the shape of the skin surface obtained by the pressure detection part is acquired and the external shape of the subject is obtained. A method for controlling an electrical stimulation apparatus, characterized in that a database showing the muscle arrangement is created by recognizing a muscle. The method of controlling an electrical stimulation device according to claim 16,
(A) After having the subject take a reference posture or fixing the target part with a cast, obtain image information of the skeletal cross section and muscle or fat cross section at that time, and obtain a large number of image data with different cross-sectional positions. After building a database of static human body structures related to the subject by stacking and creating a database,
(B) Dynamically including time lapse by acquiring image data obtained by photographing a kinematic pair along the time axis while operating a joint, and creating a static database relating to the shape of the kinematic pair for each angle of the joint A database of human structures
(C) A method for controlling an electrical stimulation device, comprising creating a database indicating a muscle arrangement of a subject from the database obtained in (b). The method of controlling an electrical stimulation device according to claim 19,
A method of controlling an electrical stimulation device, characterized in that, when identifying a muscle in a tomographic section, the color difference between a lubricating film and a muscle between the muscles is identified. The method of controlling an electrical stimulation device according to claim 19,
In creating the three-dimensional shape data from the contour lines of bones or muscles in each tomographic section, after calculating the surface centroid of the inner area of the contour line, A method for controlling an electrical stimulation device, wherein intersection points with the contour line are obtained and these intersection points are used as reference points for three-dimensional shape data. The method of controlling an electrical stimulation device according to claim 19,
When expressing muscle motion with polygon data, the spline curve that passes through the center of gravity of each tomographic plane is used as the center line, and the polygon data of the muscle is changed over time according to the shape of the center line. A method for controlling an electrical stimulation device, comprising: The method of controlling an electrical stimulation device according to claim 20,
Electrical stimulation characterized by estimating the response speed and endurance of the subject by identifying the difference in color of the muscle distribution in the cross-section and determining the quantitative ratio of the red and white muscles Control method of the device. The method of controlling an electrical stimulation device according to claim 19,
After setting the attenuation rate for each material of bone, muscle, fat, and tendon, and finding a place where the energy loss to each part of the skin surface is small, an electrode for interference wave electrical stimulation is placed at that place. A method for controlling an electrical stimulation device, comprising: setting. The method of controlling an electrical stimulation device according to claim 19,
Set material density for each material for bones, muscles, fats, and tendons, find a place where the phase shift to each part of the skin surface is small, and then apply interference wave electrical stimulation to that place. A method for controlling an electrical stimulation device, comprising setting an electrode. The method of controlling an electrical stimulation device according to claim 19,
(A) After setting the wave attenuation rate, material density, reflectance and refractive index for each material for bone, muscle, fat and tendon,
(B) Based on the attenuation rate or substance density set in (a), a location where the energy loss from the skin surface to the target site is small or a location where the phase shift is small is obtained, and a plurality of electrode candidates at that location are obtained. Select the electrode of
(C) The frequency of the interference wave on the plane is simulated by planarly simulating the waveform reflected or refracted at the part with respect to the temporal transition of the carrier wave from the electrode position selected in (b) to the target part. Find the distribution
(D) A method of controlling an electrical stimulation device, wherein the steps (b) and (c) are repeated until an ideal interference wave at a target site is obtained. The method of controlling an electrical stimulation device according to claim 26,
A control method for an electrical stimulation apparatus, comprising: setting an output of stimulation intensity to a target site of electrical stimulation by changing an output current value of an electrode finally selected from electrode candidates. The method of controlling an electrical stimulation device according to claim 26,
After acquiring a plurality of electrode candidates, the position information is stored, and when actually used as an electrode, the electrode candidate position information is read out, and an electrode suitable for the target region of electrical stimulation is selected from these. A method for controlling an electrical stimulation device, comprising: selecting an electrical stimulation device. The method of controlling an electrical stimulation device according to claim 26,
Among the electrode candidates, the myoelectric potential information is detected from the same position as the finally selected electrode position, and the muscle tonus is predicted by performing frequency analysis on each signal from the anode side electrode and the cathode side electrode. Method for controlling an electrical stimulator. The method of controlling an electrical stimulation device according to claim 26,
Control of an electrical stimulator characterized by preliminarily grasping muscle movement from muscle tonus predicted from frequency analysis and predicting the next movement of the subject's joint using a dynamic human body structure database Method. The method of controlling an electrical stimulation device according to claim 29,
A method of controlling an electrical stimulation device, wherein time-division processing is performed so that the application of electrical stimulation through a selected electrode and the processing for acquiring myoelectric potential information using the electrode as a detection electrode are not performed simultaneously .

Images