200941236 九、發明說明： 【發明所屬之技術領域】 本發明係有關於一種「仿真智慧型機器人之模擬教導系統」，尤指 經由該模擬系統教導出之仿真智慧型機器人無需回饋控制，亦不需利 用伺服馬達，以直流減速馬達即能達成與真人一樣之動作與行動，大 大降低系統複雜性及成本’因而突破機器人研究上久久不能解決之兩 ®足行走穩定性問題與動作姿態控制問題，而不用姿態感測器也是本發 明之特點者。 【先前技術】 仿真智慧型服務機器人(Mimetic Intelligent Service Robot) 為狹義的機器人(Robot)，即外形、尺寸、重量、重心等像人類，動作 與行動、表情與反應模仿自然人，而本能智慧、常識與服務知識學自 0人類，並具有學習智能之人形機器人(Humanoid Robot)。此種仿真智 慧型機器人所用到的軟體技術層次及複雜度均高，但硬體研發及量產 並不困難，因為關鍵元組件均利用現有的市售商品，組裝與整合測試 技術也都有現成可用，唯有動作與行動控制及智慧與知識軟體需要特 別創新研發。 一般而言，目前常見的仿真智慧型機器人其所有動作控制多利用 感測模組感測機器人本身及外界之狀態’經電腦内的狀態方程式與運 動方程式計算獲得後，再驅動設於機器人各關節處的伺服馬達作動， 5 200941236 令該仿真智慧型機器人能完成簡單的基本站立、手足動作或行走，曰 因為利用感測模組經狀態方程式與運動方程式透過電腦計算^再控= 之方式，在每一個動作進行前，都需要經過一連串精確、複雜的計算， 計算過程又需要相當迅速，才能在失誤動作產生前即下達糾正控制命 令，故所應用的電腦必定要計算速度顺，記‘随夠大才能勝任，如 此來不僅會大大提南成本，在故障或例行性維修、保養時將更是 〇麻煩1讓該仿騎慧型人軸實用彳卜況且，因械體的: 動作非如-般舉手、抬_作單純’料僅要各斷算每—肢體的位 置、速度、加速度、角度、角速度等數值，還要讓各肢體能夠相搭配， 而不發生碰撞、干涉，要機器人做完全跟人—樣的動作與行動在目前 的技術上而言，仍屬尚未被突破者。 、本發明即為解決上述困境而創作，以使仿真智_機狀可在系 統複雜性及縣雜下，$齡仙者騎敝合與服務動 Ο作，在服務時還能避免肢體發生碰撞、互相干涉等現象，為本發明之 目的。 因此，若能有-種將真人的所有動作细三維(立體)動作擷取系 統予以數位化擷取編號，儲存於仿真智慧型機器人類似人類小腦的行 動控制電财_，#智慧電峨仿人_大職受人的工作命令 或自行依外在環境研判需要採取某種適當動作時，即發出代贼編號 給行動㈣咖，由其發_細人各_之命令，並由分散獨立 控制及驅動之主從控制系統驅動各關節馬達，達成機器人身體各部分 6 200941236 模仿真人之動作，乃係本發明所欲解決之問題所在。 本發明人有鑑於此，乃思及創作的意念，遂以多年的經驗加以設 計，經多方探討，乃提出此發明。 【發明内容】 解決問題之技術特點： 本發明之仿真智慧型機器人之模擬教導系統，其係將仿真智慧型 〇機器人製作成與真人相同肢_尺寸，該仿真智慧型機器人各動作部 位之重心並與真人相似，重量亦相似，再由相近似身材之教導人員在 三維動作擷取系統(3D Motion Capture System)擷取範圍内演示各種 行動及動作，由三_作獅_將真人崎__猶為數位資 料，並與由智慧電腦所編輯好的服務動作代碼存入仿真智慧型機器人 行動控制賴騎作龍庫中_用’而仿真智慧型機狀各關節的 驅動馬達係採用直流馬達，其扭矩與轉速依教導人員身體各部位動作 〇之迷度與力量予以匹配設計適當之減速比，各_馬達之轉動扭力與 轉速由行動·賴自祕財抑，直接轉縣轉速與轉數之 數據’透過單晶片微處理H與驅動器直接驅動機器人之各卿馬達。 且因為所有動作均經由教導人員動作得到之動作數據，即由真人小腦 之動作平衡直接由機器人之電腦透過三維動作娜系統學習，並據以 使機器人錄，科錄概算，核f知道機狀讀各部之狀鎮 =程式、運動方程式與身驅平衡控制式；藉此，係能構成該仿真 曰慧型機器人之模擬教導系統者。 200941236 對照先前技術之功效： 本發狀仿真智慧賴n人之模擬教導祕即糊完全模仿人 類體型、尺寸、各部重量及重心之仿真智慧型機器人與真人會有相同 之動作，而借用人類小腦所儲存的運動穩定控制之經驗，以類似專家 系統之方法，完全避開傳統兩足機器人利用控制理論繁複計算之行走 及動作穩疋控_題；在教導仿真智慧型㈣人時係用經訓練之機 ❾器人教導人員示範動作，經三維動作擷取系統將肢體的各部動作轉化 為各關料狀轉速與轉數㈣，独代碼__祕存在機胃人之行動 控制電版巾’在重演時貞，丨自行動控制電腦之動作資料庫依動作代碼取 出動作數據’透過由單晶片微處理器及驅動器/馬達組成之控繼驅 動各獨立之關節馬達同時動作，以進行合成動作及複雜之服務動作者。 【實施方式】 為使貴審查委員對本發明之目的、特徵及功效能夠有更進一步 〇之瞭解與認識，以下茲請配合【圖式簡單說明】詳述如后： 通常根據本發明，先由第-圖搭§&第二騎示觀之，其係將仿真 智慧型機器人⑻做雜真人的尺寸，該仿真智慧型齡人⑻各動作 部位之重心（質量中心)並與真人相似，重量亦相似，再由相近似身材 之教導人員（P)在三維（立體)動作擷取系統(10)擷取範圍内演示各種 行動及動作’由三維動作擷取系統(10)將行動與動作轉換為數位資 料，並與由機器人之智慧電腦(20)所編輯好的服務動作代碼存入仿真 智慧型機器人(R)行動控制電腦(30)的動作資料庫(31)中内備用，或能 200941236 以實時(Real Time)方式直接自三維動作擷取系統⑽取得各關節動 作角度及角速度’並立刻傳送到仿真智慧型機器人⑻，不經動作資料 庫(31)即能與教導人員（p)同步動作，實現實時教導重演之任何動作， 又仿真智慧型機器人(R)各關節的驅動馬達係採用直流馬達（亦可為直 流概馬達），其扭矩與轉速依教導人員⑻身體各部位動作之速度與 力量予以匹配設計適當之減速比。各關節馬達之轉速與轉數由行動控 ❹制電腦(30)自動作資料庫⑻中取出，將此數據，透過單晶片微處理 盗（40)與驅動器（5〇)直接驅動各關節馬達，因為仿真智慧型機器人⑻ 的所有動作均經由教導人員（p)教示動作得到之動作數據，即由人腦之 動作平衡直接由電腦透過三維動作練取系統⑽學習，並據以重演 (Playback) ’故不需控制運算，也無需知道機器人身體各部之運動方 程式與身驅平衡控制運算式。 其中’二維動作擷取系統(10)為近十年來的新技術，該雖已廣泛 〇應用於電腦動畫、遊戲軟體製作、科幻虛擬、運動姿態比對、教導比 對、校正等’但尚未用於仿真智慧型機器人⑻的動作與行動教導方 面。二維動作齡系統⑽雖新興，但種類不少，本發明實施例採用 的是同時⑽光點攝賴取法，其主要共有六組，每組各二個攝取光 點移動位置之主動攝影擷取頭，能將教導人員⑻身上各處(頭、四肢、 身驅）隨人動作而移位之各光點瞬間相對位置及各關節點之位置、速 度、加速度、角度、肖速度與角加速度娜化絲錄據，這些數據 用以設計仿真智慧型機器人⑻四肢及頭部之直流馬彡或直流飼服馬 9 200941236 達減速齒輪比，以使仿真智慧型_人⑻之動倾真人在位置關節 角度與速度上完全-致’因仿真智慧型機器人⑻與真人在身體各部尺 寸、重量與重心等都一致者β 再者由於仿真3慧型機器人⑻的所有動作控制均非由傳統的控 制所用雜態方程式與独方料計算麟，岐軸人工演示教導 。己隐得來’所以’仿真智慧型機器人⑻的所有動作，*論是簡單的基 本站立躺臥手足動作或行走，或是複雜的組合各動作的服務動 作’甚至動作與說話的整合，以及機器人與其他機器人的協調整合， 都需要事先教導，所教導的_或組合動作，全部都由仿真智慧型機 器人⑻的智慧電腦⑽編上代碼(6至8位文字與數字組合之代碼）， 並存入仿真智慧型機ϋ人⑻_作賴庫(31)中，以便必要時叫出重 演(Playback) ’代碼即如下表所示（一部分）。 身體動作 —------—_______ 英文 動作代踢 點頭 Head Nod HD_ 〜99 搖頭 Head Wag HDWGOO〜99 張嘴/嘴開閉 Mouth Open MHONOO〜99 左臂侧舉 Left Arm Side Rise LASROO〜99 右臂侧舉 Right Ann Side Rise RAS_ 〜99 左臂前舉 Left Arm Front Rise LAF_ 〜99 200941236 右臂前舉 Right Arm Front Rise RAF_ 〜99 左肘上下彎 Left Elbow Bend Up-Down LEBU00〜99 右肘上下彎 Right Elbow Bend Up-Down REBU00〜99 左肘内外彎 Left Elbow Bend In-Out LEBI00〜99 右肘内外彎 Right Elbow Bend In-Out REBI00〜99 接著請配合第三圖所示觀之，由於本發明之仿真智慧型機器人⑻ 〇的所有動作均是模仿自真人，故與傳統機器人有下述較大之相異點： (一) 使用開口圈控制(OpenLoop Control)，沒有内回饋（internai Feedback)只有透由視覺之最終位置確認修正，即外回饋（Ex1;ernal Feedback) ° (二) 使用直流馬達取代伺服馬達，仿真智慧型機器人(R)除模仿人 眼之攝影機可上下左右轉動各45度及嘴巴之張合使用伺服馬達外，四 肢及頭部關即30個均使用直流馬達驅動，而十隻手指則由電磁器 〇 (Solenoid)予以致動。 (三） 因為目前的電池能量密度仍不夠高，為節省耗電，行走快速 及行走全時穩定起見，採用創新之滚輪溜冰拖步行走法，此法使機器 人教導人員（p)必須先學會溜冰及儀姿才能勝任教示（Teaching)仿真 智慧型機器人(R)動作、行走及服務之工作。 (四) 使用兩個完全一樣之迷你型個人電腦(Minipc)，其一為模仿 人類大腦之智慧電腦（11^11丨2611(；^{：〇1^1^€『)（2〇)，另一為模仿人類 小腦之行動控制電腦以〇乜〇11(：〇115)1^6]^(30)，兩者均為具使用雙核心 11 200941236 中央處理器（Duo Core CPU)之高速運算(2 GHz)型’且有80G之記憶容 量，其輸出各有6個RS232及4個USB埠，另有802.11、IEEE3914及 藍芽等無線通訊卡及介面。 (五）以本發明所述之創新教導重演方法來快速教會一部仿真智慧 型機器人(R)所有我們要它做的動作及行動，所有其他機器人即可複製 動作數據資訊’自動學會所有經教示之動作。 ❹ （六）利用超視覺(Super Vision)取代超音波、雷射或紅外線作防 撞及距離感測’系統簡單化’也不用陀螺儀、水平傾斜感測器等來作 為身體重心平衡控制之回饋輸入，除人類内需的感覺器官如鼻子之嗅 覺、皮膚之觸覺、舌頭之味覺等外，其他外顯功能之器官如眼睛之視 覺以數位攝影機模仿，耳朵之聽覺以拾音器（Micr〇ph〇ne)模仿，嘴巴 之動作及發聲由揚聲器（Speaker)及下巴動作伺服馬達予以模仿。 (七） 此仿真智慧型機器人(R)具有設計置入(Design_in)之機器人 〇饤為道德，且具有人工辅助之是非、善惡及慈悲等之學習能力。 (八) 具防抄襲保護及自毀本能智慧及服^務動作與基本動作資料庫 等軟體之能力，所有通訊均加密以防侧錄抄襲，並由顧客服務中心之 飼服器予以保全(Save)。 據此’當人需要仿真智慧型機器人⑻服務或機器人自己依據視覺 與聽覺感狼内辦鐘與電力監控衫方面資訊满需要行動或動作 時’它將進入重演(Playback)步驟(請同時由第四圖所示觀之） ，此時 仿真智慧型機器人⑻之所以能夠完全模仿教導人員⑻之動作，關鍵 12 200941236 在於叹°十動作控制系統時，各可同時獨立運動之關節（機器人全身共有 直/瓜馬達、伺服馬達及電磁器等45個致動器）均可獨立驅動，依邪動 作擷取之實體運動角度與角速度數據換算為馬達之轉數與轉速，故當 自智慧電腦⑽)獲得動作或服務代碼(Service Code)時，行動控制電 腦（30)即自其動作資料庫(31)取出已教示之動作數據，轉換成各馬達 之轉速與轉數，並以相同之時基一起動作，而構成機器人 ❾之合成動作’如果是服務用的連續動作，則依所教示之節目（pr〇驭咖） 一幕一幕叫出動作數據重演’及行動控制電腦(30)的控制命令是透過 6組8051系列單晶片當控制發號中心(Hub)，再將控制信號分送(每組 最多12個）各驅動器（5〇)之單晶片微處理器(4〇)，以控制通往馬達或 驅動器（50)之電流，所有動作速度控制之電流大小皆採脈衝寬度調制 (Pulse Width Modulation,PWM)，因真人（自然人）的動作在軌跡及速 度上不是很精碟，也就是重複出現性差，故此種各關節馬達之速度分 ◎級已經足夠模仿人的所有動作，而使重演順利。 至於動作穩定之控制問題，因仿真智慧型機器人(R)模仿教導人員 (P)的小腦控制肢體的方式，完全避開繁複之狀態方程式、運動方程式 及穩定性控制運算式的繁複計算，完全以叫出資料庠中之動作數據方 式以開口圈（Open Loop)控制鏈之方式進行(請同時由第五圖所示觀 之），故在動作穩定之控制上將更簡單、有效、低成本且可靠者。 另，因為人體關節數眾多’故控制上亦可採取分群控制方式，例 如行動相關（腿部以下)與動作相關（上半身）由兩個單晶片分派動作命 13 200941236 令，再由各自獨立之關節軸馬達(也是由單晶片控制）予以控制轉速與 轉數，其動作基點(Home)及終點(極限）則各有極限開關(LimitSwitch) 予以限定其行程。 藉由上述之系統，即利用完全模仿人類體型、尺寸、各部重量及 重心之仿真智慧型齡人⑻與真人會有_之勒，而翻人類小腦 所儲存運動狀控制之經财法’靖似行料家纽枝，完全避 〇開傳統控制理論繁複之兩足機器人的行走及動作穩定控制問題。在教 導仿真智慧型機器人(R)時’係用經訓練之機器人教導人員（p)示範動 作，經二維動作擷取系統(1〇)將肢體的各部動作轉化為各關節馬達之 轉速與轉數數據，並以代碼一起儲存在行動控制電腦（3〇)中，在重演 時則自行動控制電腦（30)之動作資料庫（31)依動作代碼取出動作數 據，透過由單晶片微處理器(4〇)及驅動器（50)/馬達組成之控制鍵，驅 動各獨立之關節馬達同時動作，以進行合成動作及複雜之服務動作， ❹經實驗證實本發明簡單、有效、低成本、且可靠者。 唯以上所述者，僅為本發明之一較佳實施例而已，當不能以之限 定本發明實施之範圍；即大凡依本發明申請專利範圍所作之均等變化 與修飾，皆應仍屬本創作專利涵蓋之範圍内。 200941236 【圖式簡單說明】 第一圖：係本發明動作擷取之示意圖。 第二圖：係本發明之擷取動作方塊流程圖。 第三圖：係本發明仿真智慧型機器人之系統架構圖。 第四圖：係本發明仿真智慧型機器人動作重演之流程圖。 第五圖：係本發明仿真智慧型機器人之動作控制鏈圖。 【主要元件符號說明】200941236 IX. Description of the Invention: [Technical Field of the Invention] The present invention relates to a "simulation teaching system for simulating a smart robot", in particular, a simulated intelligent robot taught by the simulation system does not require feedback control, and does not need By using the servo motor, the DC gear motor can achieve the same action and action as the real person, greatly reducing the system complexity and cost. Thus, it breaks through the two-foot walking stability problem and the motion attitude control problem that the robot research can not solve for a long time. The absence of an attitude sensor is also a feature of the present invention. [Prior Art] Mimetic Intelligent Service Robot is a narrow robot, that is, shape, size, weight, center of gravity, etc. Like human beings, movements and actions, expressions and reactions imitate natural persons, and instinct wisdom and common sense Learn from the knowledge of 0 human beings, and have a humanoid robot (Humanoid Robot). The software technology and complexity used in this kind of simulation smart robot are high, but hardware development and mass production are not difficult, because the key components are all using existing commercial products, and the assembly and integration test technologies are also readily available. Available, only action and action control and wisdom and knowledge software require special innovation and research and development. Generally speaking, the current common simulation intelligent robot uses all the motion control to sense the state of the robot itself and the outside world. After being calculated by the equation of state and the equation of motion in the computer, it is driven to the joints of the robot. The servo motor is activated, 5 200941236 enables the simulated intelligent robot to perform simple basic standing, hand and foot movements or walking, because the sensing module is calculated by the state equation and the motion equation through computer calculation ^ recontrol = Before each action, a series of precise and complicated calculations are required, and the calculation process needs to be quite rapid, so that the corrective control command can be issued before the error action occurs. Therefore, the applied computer must calculate the speed and read, The ability to do so will not only greatly increase the cost of the South, but will also be more troublesome in the event of fault or routine maintenance and maintenance. 1 Let the imitation rider's human axis be practical and difficult, and because of the mechanical action: - Raise your hand, lift _ as a simple material, only need to calculate each position, speed, acceleration, angle, Numerical speed, but also to each limb can be matched without collision, interference with people to do exactly the robot - like movements and actions in the present Technically, who still has not been a breakthrough. The present invention is created to solve the above-mentioned predicament, so that the simulation wisdom_machine can be used in the complexity of the system and the county, the age-old rider and the service are active, and the limb collision can be avoided during the service. The phenomenon of mutual interference and the like is the object of the present invention. Therefore, if there is a kind of three-dimensional (stereo) motion capture system that takes all the actions of the real person, the number is digitized and retrieved, and the simulation intelligent robot is similar to the human cerebellum's action control power _, #智慧电峨人_When a job order of a large-time job is required to take some appropriate action according to the external environment, the thief number is issued to the action (4) coffee, and the order of the _ _ _ _ _ _ _ _ _ _ _ The driving master slave control system drives each joint motor to achieve the action of the various parts of the robot body, which is the problem to be solved by the present invention. In view of this, the inventor of the present invention has been thinking about the idea of creation and has been designed with many years of experience. SUMMARY OF THE INVENTION Technical Problem for Solving the Problem: The simulation teaching system of the simulated intelligent robot of the present invention is to make the simulated smart 〇 robot into the same limb size as the real person, and the center of gravity of each action part of the simulated intelligent robot Similar to real people, the weight is similar, and the instructors of the similar body demonstrate the various actions and actions in the 3D Motion Capture System. The three _ lions will be the real people __ Jude For the digital data, and the service action code edited by the smart computer is stored in the simulation intelligent robot action control Laiqiu dragon library _ use 'simulation of the intelligent machine-like joints of the drive motor is a DC motor, The torque and the speed are matched according to the fantasity and strength of the various parts of the body of the trainer. The appropriate reduction ratio is designed. The rotational torque and the rotational speed of each motor are controlled by the action, and the data of the speed and the number of revolutions are directly transferred to the county. 'The single-chip micro-processing H and the driver directly drive the individual motors of the robot. And because all the movements are obtained through the movements of the teaching staff, that is, the movement balance of the human cerebellum is directly learned by the robot's computer through the three-dimensional action system, and according to the robot, the calculation of the robot, the nuclear f knows the machine reading The town of each department = program, equation of motion and body drive balance control; thereby, it can constitute the simulation teaching system of the simulation Hui robot. 200941236 Compared with the efficacy of the prior art: This simulation simulation wisdom is based on the simulation of the human body, size, weight and center of gravity of the simulated intelligent robot and the real person will have the same action, and borrow the human cerebellum The experience of stored motion stability control, in a similar way to the expert system, completely avoids the traditional biped robot using the control theory to calculate the walking and motion stability control _ questions; teaching the simulation of intelligent (four) people with training The machine teaches the person to demonstrate the action, and the three-dimensional action capture system converts the movements of the limbs into the rotational speed and the number of revolutions of each material (4), and the unique code __ secret existence machine stomach control action electric towel' replay At the same time, the action data from the action control computer extracts the action data according to the action code. The drive is composed of a single-chip microprocessor and a driver/motor to drive the independent joint motors simultaneously for synthesizing actions and complex services. The author. [Embodiment] In order to enable the reviewing committee to have a better understanding and understanding of the purpose, features and effects of the present invention, please refer to the following [detailed description of the drawings] as follows: Generally, according to the present invention, - Figure § & second riding view, it will simulate the smart robot (8) to do the size of the real person, the simulation of the intelligent age (8) the center of gravity of each action part (quality center) and similar to the real person, the weight is also Similarly, the instructor (P) of the similar figure demonstrates various actions and actions within the scope of the three-dimensional (stereo) motion capture system (10). The three-dimensional motion capture system (10) converts actions and actions into Digital data, and the service action code edited by the robot's smart computer (20) is stored in the action database (31) of the simulated smart robot (R) action control computer (30), or can be used in 200941236 The real time (Real Time) method directly acquires the joint angle and angular velocity of each joint from the 3D motion capture system (10) and immediately transmits it to the simulated intelligent robot (8), and can teach the person without the motion database (31). (p) Synchronous action, realizing any action of real-time teaching replay, and simulating the driving motor of each joint of the intelligent robot (R) adopts DC motor (also can be DC motor), its torque and speed are according to the teaching staff (8) The speed and strength of the part action are matched to the appropriate reduction ratio. The speed and number of revolutions of each joint motor are automatically taken out by the mobile control computer (30), and the data is directly driven by the single-chip micro-processing thief (40) and the driver (5 〇). Because all the actions of the simulated smart robot (8) are obtained through the teaching staff (p) to teach the motion data, that is, the balance of the human brain is directly learned by the computer through the three-dimensional motion learning system (10), and according to the replay (Playback) ' Therefore, there is no need to control the operation, and it is not necessary to know the motion equations of the various parts of the robot body and the body drive balance control arithmetic formula. Among them, the 'two-dimensional motion capture system (10) is a new technology in the past ten years, which has been widely used in computer animation, game software production, science fiction virtual, motion pose comparison, teaching comparison, correction, etc. 'but not yet Used to simulate the action and action teaching aspects of the intelligent robot (8). The two-dimensional action age system (10) is new, but there are many types. The embodiment of the present invention adopts the simultaneous (10) light spot detection method, which mainly has six groups, and each group has two active photo capture positions for taking light spot moving positions. The head, the relative position of each light spot that can be displaced by the teaching staff (8) (the head, the limbs, and the body drive) and the position, velocity, acceleration, angle, Xiao speed and angular acceleration of each joint point According to the silk recording data, these data are used to design the intelligent robot (8) DC and horse heads of the limbs and the head of the horse, and the speed of the reduction gear ratio, so that the simulation intelligence type _ people (8) move the real person in the position joint The angle and speed are completely-induced. Because the simulated intelligent robot (8) and the real person are consistent in the size, weight and center of gravity of the body, etc., because all the motion control of the simulated 3-type robot (8) is not used by the traditional control. State equations and independent materials are used to calculate the lining, and the 岐 axis artificial demonstration teaching. It’s hidden that 'all' simulates all the actions of the intelligent robot (8), which is a simple basic standing and lying movement or walking, or a complex combination of various action service actions 'even the integration of action and speech, and the robot Coordination and integration with other robots requires prior teaching, and the _ or combined actions taught are all coded by the intelligent computer (10) of the simulated intelligent robot (8) (codes of 6 to 8 digits combined with numbers). Into the simulation smart machine (8) _ Lai library (31), so that if necessary, call the replay (Playback) 'code is as shown in the following table (part). Body movements——------________ English action generation kick nods Head Nod HD_ ~99 Shake Head Head Wag HDWGOO~99 Open mouth Mouth Open MHONOO~99 Left Arm Side Lift Left Arm Side Rise LASROO~99 Right Arm Side Lift Ann Side Rise RAS_ ~99 Left Arm Front Lift Front Arm Front Rise LAF_ ~99 200941236 Right Arm Front Lift Right Arm Front Rise RAF_ ~99 Left Elbow Bend Up-Down LEBU00~99 Right Elbow Up and Down Bend Right Elbow Bend Up-Down REBU00~99 Left Elbow Bend In-Out LEBI00~99 Right Elbow Bend In-Out REBI00~99 Next, please cooperate with the simulation wisdom of the present invention. The robot (8) 〇 all the actions are imitated from the real person, so it has the following differences with the traditional robot: (1) Using OpenLoop Control, there is no internal feedback (internai feedback) only through the visual The final position confirmation correction, ie, external feedback (Ex1; ernal Feedback) ° (2) Using a DC motor instead of a servo motor, simulating a smart robot (R) in addition to imitating the human eye Each rotatable by 45 degrees and the mouth of the outer sheets together servo motor, i.e., off limbs and head 30 are driven up and down using the DC motor, the electromagnetic device by ten square finger (Solenoid) to be actuated. (3) Because the current battery energy density is still not high enough, in order to save power consumption, fast walking and stable walking time, the innovative roller skating and walking method is adopted. This method enables the robot teaching staff (p) to learn first. Skating and posture can be used to teach the intelligent robot (R) movement, walking and service work. (4) Using two identical mini-type personal computers (Minipc), one of which is a smart computer that mimics the human brain (11^11丨2611(;^{:〇1^1^€『)(2〇), Another action control computer that mimics the human cerebellum is 〇乜〇11(:〇115)1^6]^(30), both of which are high-speed operations with dual-core 11 200941236 central processing unit (Duo Core CPU). (2 GHz) type and 80G memory capacity, the output has 6 RS232 and 4 USB ports, and 802.11, IEEE3914 and Bluetooth wireless communication cards and interfaces. (5) According to the present invention Innovative teaching reenactment method to quickly teach a simulated intelligent robot (R) all the actions and actions we want it to do, all other robots can copy the action data information 'automatically learn all the taught actions. ❹ (6) use super Super Vision replaces ultrasonic, laser or infrared light for collision avoidance and distance sensing 'system simplification' and does not use gyroscopes, horizontal tilt sensors, etc. as the feedback input of the body center of gravity balance control, except for human needs. Sensory organs such as the nose, the skin The senses of the tongue, the taste of the tongue, etc., the organs of other external functions such as the vision of the eyes are imitated by digital cameras, the hearing of the ears is mimicked by a pickup (Micr〇ph〇ne), the movements of the mouth and the sounds are acted upon by the speaker and the chin. The servo motor is imitated. (7) This simulation intelligent robot (R) has the ability to design (incorporate) the robot to be moral, and has the ability to learn artificially, right and wrong, compassion and compassion. Anti-plagiarism protection and self-destructing instinct intelligence and service capabilities and basic motion database and other software capabilities, all communications are encrypted to prevent side-view plagiarism, and is saved by the customer service center's feeding device (Save). 'When a person needs to simulate a smart robot (8) service or the robot itself needs to act or act according to the visual and auditory sense of the inner wolf and the power monitor shirt, it will enter the Playback step (please also be the fourth figure) At the moment, the simulation of the intelligent robot (8) can completely imitate the action of the instructor (8), the key 12 200941236 lies in the ten action In the system, each joint that can simultaneously move independently (the robot has a total of 45 actuators such as straight/guar motor, servo motor and electromagnetic device) can be driven independently, and the physical movement angle and angular velocity data can be converted according to the evil action. For the number of revolutions and the number of revolutions of the motor, when the action code or service code (Service Code) is obtained from the smart computer (10), the action control computer (30) takes the action data that has been taught from its action database (31) and converts it into The rotation speed and the number of revolutions of each motor are operated together with the same time base, and constitute a synthetic operation of the robot. If it is a continuous operation for service, the program is called according to the program (pr〇驭). The data replay 'and the action control computer (30) control command is to control the numbering center (Hub) through 6 groups of 8051 series single chips, and then distribute the control signals (up to 12 in each group) to each drive (5〇). Single-chip microprocessor (4〇) to control the current to the motor or driver (50), all the speed of the control speed is pulse width modulation (Pulse Width Modulation, PWM), due to real people Natural) operation is not very Prodisc on the track and speed, i.e. poor repeated, so that each such motor velocity joint points ◎ level sufficient to mimic the actions of all people, the repeat smoothly. As for the control problem of motion stability, the simulation intelligent robot (R) mimics the way the instructor (P)'s cerebellum controls the limbs, completely avoiding the complicated calculations of the complicated state equations, motion equations, and stability control algorithms. The action data method in the data 以 is performed in the form of an Open Loop control chain (please view it as shown in the fifth figure), so it will be simpler, more efficient, and less costly in controlling the stability of the action. Reliable. In addition, because there are a large number of joints in the human body, it is also possible to control the group by means of group control. For example, the action-related (below the legs) and the action-related (upper body) are assigned by two single-chips, and the independent joints are used. The shaft motor (also controlled by a single chip) controls the speed and the number of revolutions. The home base and the end point (limit) each have a limit switch (LimitSwitch) to limit the stroke. Through the above system, that is, using the simulation of intelligent human body (8) that completely mimics the human body shape, size, weight and center of gravity, and the real person will have a leap, while turning the human cerebellum to store the movement control of the financial law 'jinglike The stalker's new branch completely avoids the problem of walking and motion stability control of the biped robot with the traditional control theory. When teaching the simulation of the intelligent robot (R), the trained robot teaching staff (p) demonstrates the action, and the two-dimensional motion capture system (1〇) converts the motions of the limbs into the rotational speed and rotation of each joint motor. The data is stored in the action control computer (3〇) together with the code. In the replay, the action data (31) from the action control computer (30) extracts the action data according to the action code through the single-chip microprocessor. The control button composed of (4〇) and the driver (50)/motor drives the independent joint motors to simultaneously perform the synthetic action and the complicated service action. The experiment proves that the invention is simple, effective, low-cost and reliable. By. The above is only a preferred embodiment of the present invention, and the scope of the present invention is not limited thereto; that is, the equal variation and modification of the scope of the patent application of the present invention should still belong to the present creation. Within the scope of the patent. 200941236 [Simple description of the diagram] The first figure is a schematic diagram of the action capture of the present invention. The second figure is a flow chart of the capture action block of the present invention. The third figure is a system architecture diagram of the simulated intelligent robot of the present invention. The fourth figure is a flow chart of the replay of the simulated intelligent robot in the present invention. Fig. 5 is a motion control chain diagram of the simulated intelligent robot of the present invention. [Main component symbol description]
本發明部份： 仿真智慧型機器人一（R) 教導人員-----（P) 三維動作擷取系統一（10) 智慧電腦-----（20) 行動控制電腦---（30) 動作資料庫----(31) 單晶片微處理器一_(40) 驅動器------（50) 15Part of the Invention: Simulation Intelligent Robot One (R) Instructor-----(P) Three-Dimensional Motion Capture System One (10) Smart Computer-----(20) Mobile Control Computer---(30 Action Database----(31) Single Chip Microprocessor One_(40) Driver ------(50) 15