JP2021189508A

JP2021189508A - Robot, moving path generation device and program thereof, and movement prediction device

Info

Publication number: JP2021189508A
Application number: JP2020091089A
Authority: JP
Inventors: 允啓亀▲崎▼; Mitsuhiro Kamezaki; 優佑円谷; Yusuke Tsuburaya; 太智金田; Taichi Kanada; 重樹菅野; Shigeki Sugano
Original assignee: Waseda University
Current assignee: Waseda University
Priority date: 2020-05-26
Filing date: 2020-05-26
Publication date: 2021-12-13
Anticipated expiration: 2040-05-26
Also published as: JP7490193B2

Abstract

To generate an optimum moving path while considering a moving efficiency of the robot and loads on persons present in the periphery, by constructing a movement prediction model including persons or the like around a robot autonomously moving in a crowded environment.SOLUTION: A robot 10 includes a detection device 12 for detecting environment information of obstacles around the robot, and a control device 13 for controlling the movement of the robot 10 on the basis of the detection result of the detection device 12. The control device 13 includes candidate path retrieval means 24 which generates a plurality of candidate paths on the basis of position and speed information of the obstacles, and optimum path extraction means 25 which extracts an optimum path being optimum among the candidate paths. The optimum path extraction means 25 includes a movement prediction simulation unit 28 which simulates temporal moving states of the robot 10 and the moving obstacles according to the movement of the robot 10 and the interaction of the obstacles for each of the candidate paths, and a final determination unit 29 which determines the final path on the basis of the moving efficiency of the robot 10 and influences exerted on the moving obstacles according to the simulation result.SELECTED DRAWING: Figure 1

Description

本発明は、多数の人間や物体等の障害物が存在する混雑環境下においても、適切な自律移動を可能にするロボットと、当該ロボットの移動経路を生成するための移動経路生成装置及びそのプログラム、並びに、移動予測装置に関する。 The present invention is a robot that enables appropriate autonomous movement even in a congested environment where many obstacles such as humans and objects exist, a movement route generation device for generating the movement route of the robot, and a program thereof. , And the movement prediction device.

近時において、人間との共存環境下で自律的に移動可能となるロボットが出現している。当該ロボットとしては、所定のスタート地点からゴール地点まで自律移動する際に、周囲の人間を含む障害物の状況をセンシングし、当該障害物との将来的な干渉を回避しながら移動するものがある。このようなロボットにおいては、単に一方的に人間を避けるだけではなく、人間の動きや他の障害物等に配慮し、干渉を効果的に避けるための人間との協調的な動作が要求される。すなわち、当該環境下での移動時には、干渉回避動作や停止動作等、人間の動作に対して受動的な動作に加え、当該動作を通じてロボットの行動意図を人間に伝達し、人間自身の干渉回避行動を能動的に働きかけることが必要となる。例えば、ロボットの移動経路上に、ロボットがすり抜けられない程の隙間を介して多くの人間が密集しているような場合に、当該隙間に向ってロボットを移動させつつも、ロボットの通路を確保するために、前記隙間の周囲の人間に移動を促す必要が生じる。 Recently, robots that can move autonomously in a coexisting environment with humans have appeared. Some robots move autonomously from a predetermined start point to a goal point while sensing the situation of obstacles including surrounding humans and avoiding future interference with the obstacles. .. In such robots, it is required not only to avoid humans unilaterally, but also to consider human movements and other obstacles, and to cooperate with humans to effectively avoid interference. .. That is, when moving in the environment, in addition to passive movements such as interference avoidance movements and stop movements, the robot's action intention is transmitted to humans through the movements, and the human own interference avoidance behaviors. It is necessary to actively work on. For example, when many people are crowded on the robot's movement path through a gap that the robot cannot slip through, the robot's passage is secured while moving the robot toward the gap. In order to do so, it becomes necessary to encourage people around the gap to move.

そこで、本発明者らは、自身の研究成果として、前述の働きかけを踏まえた経路を計画して移動するロボットを種々提案している（特許文献１、２等参照）。特に、特許文献２においては、人間との干渉が予測される場合に、人間からある程度の空間を隔ててすれ違う移動のみならず、状況に応じて人間に接近若しくは接触しながら移動する協調移動手法が開示されている。 Therefore, as a result of their own research, the present inventors have proposed various robots that plan and move a route based on the above-mentioned action (see Patent Documents 1, 2, etc.). In particular, in Patent Document 2, when interference with a human is predicted, not only the movement of passing by a certain space from the human but also the coordinated movement method of moving while approaching or contacting the human depending on the situation is used. It has been disclosed.

また、特許文献３には、人間とすれ違う際に人間に与える不安感を低減しつつ、より短い移動距離で目的地に移動する自律移動装置が開示されている。この自律移動装置では、周囲の環境に応じて複数の移動経路を計算し、各移動経路について、それらの長さに対応する動作効率の順位付けを行う移動経路計算部と、動作効率の順位に従って移動候補を選択する移動候補選択部と、移動候補選択部で選択された移動候補について周囲の人間に不安感を与えるか否かを判断する斥力判定部と、人間に不安感を与えない最も動作効率の良い順位の移動候補を移動経路とする誘導動作指示部とを備えている。前記斥力判定部では、自律移動装置から人間に与える仮想的な斥力を算出し、当該斥力が大きい程、人間に不安感を与えるとして、予め設定された斥力の閾値を基準に不安感の有無が判定される。ここでの斥力の算出には、公知のＳＦＭ（ＳｏｃｉａｌＦｏｒｃｅＭｏｄｅｌ）が用いられる。このＳＦＭは、歩行者等の移動可能なエージェントが複数存在する環境中に仮想的な力を導入し、エージェントに対する目的地からの引力と、静止状態の壁等の障害物であるオブジェクトや他のエージェントからの斥力とをそれぞれ合成することにより、エージェントの移動を予測するモデルである。 Further, Patent Document 3 discloses an autonomous mobile device that moves to a destination with a shorter travel distance while reducing anxiety given to the human when passing by the human. In this autonomous mobile device, a movement route calculation unit that calculates a plurality of movement routes according to the surrounding environment and ranks the operation efficiency corresponding to their lengths for each movement route, and a movement route calculation unit according to the order of the operation efficiency. A movement candidate selection unit that selects a movement candidate, a repulsive force determination unit that determines whether or not the movement candidate selected by the movement candidate selection unit gives anxiety to surrounding people, and the most motion that does not give anxiety to humans. It is equipped with a guidance operation instruction unit that uses a movement candidate in an efficient order as a movement route. The repulsive force determination unit calculates a virtual repulsive force given to a human from an autonomous moving device, and considers that the larger the repulsive force is, the more anxious the human is, and whether or not there is anxiety based on a preset threshold value of the repulsive force. It is judged. A known SFM (Social Force Model) is used for the calculation of the repulsive force here. This SFM introduces a virtual force in an environment where there are multiple movable agents such as pedestrians, attracting the agent from the destination, objects that are obstacles such as stationary walls, and other objects. It is a model that predicts the movement of the agent by synthesizing the repulsive force from the agent.

特開２０１９−８４６４１号公報Japanese Unexamined Patent Publication No. 2019-84641 特開２０２０−４６７５９号公報Japanese Unexamined Patent Publication No. 2020-46759 特許第５５３９５９６号公報Japanese Patent No. 5539596

人間が多く存在する混雑環境下でロボットを自律的に移動させる場合、密接した多くの人間に接近した状態となることから、ロボットの移動に伴って移動する人間と、当該人間の移動に伴って移動する別の人間の行動を配慮したロボットの移動計画が必要となる。前記各特許文献では、前記混雑環境下において、予め定めた経路に従ってロボットが移動した際に、影響を及ぼす周囲の複数の人間への移動伝播は考慮されていない。当該移動伝播の考慮に際して、各人間の移動予測が必要になるが、当該移動予測としては、前述のＳＦＭを用いた手法がある。しかしながら、従来のＳＦＭでは、モデル内で移動可能なエージェントの認知状態について考慮されていない。例えば、対象エージェントの後方から接近する他のエージェントからの斥力も対象エージェントの移動予測に影響してしまうが、実空間では、人間が後方の人間を認知して避ける行動は不自然である等、エージェントの移動予測をより正確に行えない。また、従来のＳＦＭでは、特許文献２に開示されているようにロボットが人間に接触しながら人間の移動を促すような場合について、その接触力が考慮されないため、当該接触力の影響を反映したエージェントの移動シミュレーションができない。 When a robot is autonomously moved in a crowded environment where many humans are present, it is in a state of being close to many humans in close contact with each other. It is necessary to plan the movement of the robot in consideration of the behavior of another person who moves. In each of the patent documents, movement propagation to a plurality of surrounding human beings, which affects when the robot moves according to a predetermined path in the crowded environment, is not considered. In considering the movement propagation, it is necessary to predict the movement of each person, and as the movement prediction, there is a method using the above-mentioned SFM. However, conventional SFM does not consider the cognitive state of agents that can move within the model. For example, repulsive force from other agents approaching from behind the target agent also affects the movement prediction of the target agent, but in real space, it is unnatural for humans to recognize and avoid humans behind. It is not possible to predict the movement of agents more accurately. Further, in the conventional SFM, when the robot is in contact with a human and promotes the movement of the human as disclosed in Patent Document 2, the contact force is not taken into consideration, so that the influence of the contact force is reflected. Agent movement simulation is not possible.

本発明は、このような課題を解決するために案出されたものであり、その目的は、混雑環境下で自律移動するロボットの経路生成に適した周囲の人間等を含む移動予測モデルを構築し、ロボットの移動効率と周囲の人間等に与える負担を考慮しながら、最適な移動経路を生成することができるロボット、移動経路生成装置及びそのプログラム、並びに、移動予測装置を提供することにある。 The present invention has been devised to solve such a problem, and the purpose of the present invention is to construct a movement prediction model including surrounding human beings suitable for path generation of a robot that autonomously moves in a congested environment. The present invention is to provide a robot, a movement route generator and its program, and a movement prediction device capable of generating an optimum movement route while considering the movement efficiency of the robot and the burden on surrounding human beings. ..

前記目的を達成するため、本発明は、主として、周囲に存在する障害物を避けながら自律移動するロボットにおいて、前記障害物の位置情報及び速度情報を検出する検出装置と、当該検出装置の検出結果から、前記障害物の状況を考慮しながら前記自律移動を制御する制御装置とを備え、前記制御装置は、前記障害物の位置情報及び速度情報に基づいて、前記ロボットと前記障害物のうち移動可能な移動障害物とからなるエージェントのうち対象となる対象エージェントが移動する経路の候補となる候補経路を複数生成する候補経路探索手段と、前記各候補経路の中から最適となる最適経路を抽出する最適経路抽出手段と、当該最適経路に基づいて前記自律移動の動作指令をする動作指令手段とを備え、前記最適経路抽出手段は、前記各候補経路について、前記ロボットと前記障害物の相互作用により、前記各エージェントの移動状態をシミュレーションする移動予測シミュレーション部と、当該移動予測シミュレーション部でのシミュレーション結果から、前記対象エージェントの移動効率及び前記移動障害物に与える影響に基づいて前記最適経路を決定する最終決定部とを備え、前記移動予測シミュレーション部では、前記各エージェント間や当該各エージェントと常時静止状態の前記障害物であるオブジェクトとの間で相互に作用する仮想的な力を利用して、前記各エージェントの位置及び速度の変化が経時的に予測され、前記最終決定部では、前記各候補経路について、予め記憶された数式により、前記移動効率に関する第１の指標値と、前記移動障害物に与える負荷に関する第２の指標値とが求められ、これら各指標値を総合して得られる経路コストの最も少ない前記候補経路が前記最適経路として選択される、という構成を採っている。 In order to achieve the above object, the present invention mainly comprises a detection device that detects position information and speed information of the obstacle in a robot that autonomously moves while avoiding obstacles existing in the surroundings, and a detection result of the detection device. Therefore, the control device for controlling the autonomous movement while considering the situation of the obstacle is provided, and the control device moves among the robot and the obstacle based on the position information and the speed information of the obstacle. Candidate route search means for generating a plurality of candidate routes that are candidates for the route to which the target agent moves among agents consisting of possible movement obstacles, and extraction of the optimum route from each of the candidate routes. The optimum route extraction means includes an operation command means for issuing an operation command for the autonomous movement based on the optimum route, and the optimum route extraction means interacts with the robot and the obstacle for each candidate route. From the simulation results of the movement prediction simulation unit that simulates the movement state of each agent and the movement prediction simulation unit, the optimum route is determined based on the movement efficiency of the target agent and the influence on the movement obstacle. The movement prediction simulation unit is provided with a final determination unit, and the movement prediction simulation unit utilizes a virtual force that interacts between the agents or between the agents and the obstacle object that is always stationary. , Changes in the position and speed of each agent are predicted over time, and in the final determination unit, the first index value regarding the movement efficiency and the movement obstacle are obtained by the mathematical formula stored in advance for each candidate route. A second index value relating to a load applied to an object is obtained, and the candidate route having the lowest route cost obtained by summing up these index values is selected as the optimum route.

本発明によれば、混雑環境下で自律移動するロボットの経路生成に適した周囲の人間等を含む移動予測モデルに基づき、ロボットの移動効率と、周囲の人間等に与える物理的負担及び心理的負担からなる移動負荷とを総合的に判断することにより、ロボットと周囲の人間との相互作用をより正確に考慮した最適な移動経路を生成可能となる。つまり、ロボットの移動効率と、ロボットの移動に伴う人間の回避行動時の移動負荷とのバランスを取りながら、複数の候補経路から最適経路を選択することができる。例えば、周囲の人間等の移動負荷が多少高くても、ロボットの移動効率が格段に良ければ、その際の候補経路が最適経路として選択され、逆に、ロボットの移動効率が多少高くても、人間等の移動負荷が格段に低ければ、その際の候補経路が最適経路として選択され得ることになる。 According to the present invention, based on a movement prediction model including surrounding humans and the like suitable for path generation of a robot that autonomously moves in a congested environment, the movement efficiency of the robot and the physical burden and psychological burden on the surrounding humans and the like are given. By comprehensively judging the movement load consisting of the load, it is possible to generate an optimum movement route that more accurately considers the interaction between the robot and the surrounding human beings. That is, it is possible to select the optimum route from a plurality of candidate routes while balancing the movement efficiency of the robot and the movement load during the human avoidance action accompanying the movement of the robot. For example, even if the movement load of surrounding humans is a little high, if the movement efficiency of the robot is remarkably good, the candidate route at that time is selected as the optimum route, and conversely, even if the movement efficiency of the robot is a little high, If the movement load of humans or the like is extremely low, the candidate route at that time can be selected as the optimum route.

また、本発明における移動予測モデルは、ロボットが人間に接触しながら移動する際に、その接触力が考慮されるとともに、ロボットの周囲の人間等の認知範囲を考慮して斥力の作用状態を変化させることができるため、ロボットが周囲の人間に接触しながら混雑環境下を移動する場合に、ロボットや人間等の各エージェントについて、より適切な移動予測が可能になる。 Further, in the movement prediction model of the present invention, when the robot moves while touching a human, the contact force is taken into consideration, and the action state of the repulsive force is changed in consideration of the cognitive range of humans and the like around the robot. Therefore, when the robot moves in a crowded environment while in contact with surrounding humans, it becomes possible to more appropriately predict the movement of each agent such as a robot or a human.

本実施形態に係るロボット１０の移動に関連する構成のみを概略的に表したブロック図である。It is a block diagram which represented only the structure which concerns on the movement of the robot 10 which concerns on this embodiment. 候補経路探索手段での候補経路の生成を説明するための概念図である。It is a conceptual diagram for demonstrating the generation of a candidate route by a candidate route search means. （Ａ）〜（Ｆ）は。移動予測シミュレーション部での各エージェントの移動予測シミュレーションを時系列で説明するための概念図である。(A) to (F) are. It is a conceptual diagram for explaining the movement prediction simulation of each agent in the movement prediction simulation part in time series.

以下、本発明の実施形態について図面を参照しながら説明する。 Hereinafter, embodiments of the present invention will be described with reference to the drawings.

図１には、本実施形態に係るロボット１０の移動に関連する構成のみを概略的に表したブロック図が示されている。この図において、前記ロボット１０は、予め設定された移動開始地点（スタート地点）から所定の目標地点（ゴール地点）まで自律的に移動する移動体として機能する。このロボット１０では、移動途中で壁等の物体や人間等からなる障害物との将来的な干渉が予想される場合に、これら障害物を避けながら進行する移動経路として、周囲の人間等への影響を軽減して効率良く移動可能な移動経路を探索し、当該移動経路に沿った動作制御がなされる。 FIG. 1 shows a block diagram schematically showing only the configuration related to the movement of the robot 10 according to the present embodiment. In this figure, the robot 10 functions as a moving body that autonomously moves from a preset movement start point (start point) to a predetermined target point (goal point). In this robot 10, when future interference with an object such as a wall or an obstacle consisting of human beings is expected during movement, the robot 10 travels to surrounding human beings as a moving route while avoiding these obstacles. A movement route that can be moved efficiently with reduced influence is searched for, and operation control is performed along the movement route.

ここでの障害物としては、ロボット１０の移動経路の周辺に存在する壁等、環境中で常時静止した状態の固定障害物と、人間や移動体等、環境中で移動可能となる移動障害物とがある。なお、以下においては、複数の人間を移動障害物として説明するが、他の移動体に対しても後述と同様の処理により、移動経路の生成が可能となる。 Obstacles here include fixed obstacles such as walls existing around the movement path of the robot 10 that are always stationary in the environment, and moving obstacles such as humans and mobile objects that can move in the environment. There is. In the following, a plurality of human beings will be described as movement obstacles, but it is possible to generate a movement path for other moving objects by the same processing as described later.

また、特に限定されるものではないが、本実施形態では、障害物を避けながら進行する移動経路として、次の３種の経路が対象となる。すなわち、ここでの移動経路としては、ロボット１０と人間の離間距離が所定値以上の状態を保持したまま人間との干渉を回避する回避経路と、ロボット１０と人間の離間距離が前記所定値未満であるものの、それらが接触しない接近状態ですれ違い可能にする接近経路と、ロボット１０が人間とすれ違う際に、当該人間に接触しながらロボット１０の通路を確保する働きかけを行うことを前提とした接触経路とがある。 Further, although not particularly limited, in the present embodiment, the following three types of routes are targeted as movement routes that proceed while avoiding obstacles. That is, as the movement path here, an avoidance path for avoiding interference with a human while maintaining a state where the distance between the robot 10 and the human is equal to or more than a predetermined value, and a distance between the robot 10 and the human being less than the predetermined value. However, the approach route that allows them to pass each other in an approaching state where they do not touch, and the contact on the premise that when the robot 10 passes by a human, the robot 10 works to secure the passage of the robot 10 while contacting the human. There is a route.

前記ロボット１０には、図１に示されるように、各種動作を可能に構成された機構や機器からなる動作部１１と、ロボット１０の周囲の環境情報を検出する検出装置１２と、検出装置１２の検出結果に基づき、前記障害物の状況を考慮しながらロボット１０の自律移動制御を行う制御装置１３とが設けられている。 As shown in FIG. 1, the robot 10 includes an operation unit 11 composed of mechanisms and devices configured to enable various operations, a detection device 12 for detecting environmental information around the robot 10, and a detection device 12. A control device 13 that performs autonomous movement control of the robot 10 while considering the situation of the obstacle is provided based on the detection result of the robot 10.

前記動作部１１は、所定範囲内を自律移動させるための機構やその動力源からなる移動装置１４と、所定空間内を動作するための機構やその動力源からなるアーム１５とを含む。これら移動装置１４及びアーム１５等の動作部１１については、全て公知の部材、機構、装置類等から構成されており、本発明の本質部分でないため、各構成についての詳細な図示説明を省略する。 The moving unit 11 includes a moving device 14 including a mechanism for autonomously moving within a predetermined range and a power source thereof, and an arm 15 including a mechanism for operating in a predetermined space and a power source thereof. Since the moving unit 14 and the operating unit 11 such as the arm 15 are all composed of known members, mechanisms, devices, etc. and are not an essential part of the present invention, detailed illustration and description of each configuration will be omitted. ..

前記検出装置１２は、ロボット１０の周囲に存在する人間その他の障害物の位置情報及び速度情報を検出する周囲環境検出部１８と、人間の向きを検出するための向き検出部１９とからなり、各検出部１８，１９での検出結果は、制御装置１３に逐次送信される。 The detection device 12 includes an ambient environment detection unit 18 for detecting position information and speed information of humans and other obstacles existing around the robot 10, and an orientation detection unit 19 for detecting the direction of a human. The detection results of the detection units 18 and 19 are sequentially transmitted to the control device 13.

前記周囲環境検出部１８としては、特に限定されるものではないが、ロボット１０の周囲へのレーザ光の照射による人間を含む物体の反射状態に基づいて、ロボット１０の周囲の障害物の各表面部分の位置を検出する公知のレーザレンジファインダ等の測距センサが用いられる。つまり、この周囲環境検出部１８では、各種障害物の表面部分を構成する点群における各点の平面視での２次元位置が、ロボット１０を基準として測定され、当該測定値に基づいて、前記移動障害物の速度情報が算出される。 The ambient environment detection unit 18 is not particularly limited, but is based on the reflection state of an object including a human being by irradiating the surroundings of the robot 10 with a laser beam, and each surface of an obstacle around the robot 10. A distance measuring sensor such as a known laser range finder that detects the position of a portion is used. That is, in the ambient environment detection unit 18, the two-dimensional position in the plan view of each point in the point cloud constituting the surface portion of various obstacles is measured with reference to the robot 10, and based on the measured value, the said The speed information of the moving obstacle is calculated.

なお、周囲環境検出部１８としては、ロボット１０の周囲に存在する各種障害物の位置検出等が可能な限りにおいて、ＧＰＳ等を利用したセンサ等、他のセンサや装置類を適用することも可能である。 As the ambient environment detection unit 18, other sensors and devices such as a sensor using GPS or the like can be applied as long as the position of various obstacles existing around the robot 10 can be detected. Is.

前記向き検出部１９としては、特に限定されるものではないが、ＫＩＮＥＣＴ（登録商標）等の公知のデプスセンサが用いられ、骨格認識によって人間の顔面部分が特定されるとともに、周囲環境検出部１８の検出結果から取得した人間の位置情報を利用して、ロボット１０に対する人間の顔面部分の向きが検出される。 The orientation detection unit 19 is not particularly limited, but a known depth sensor such as KINECT (registered trademark) is used, a human face portion is specified by skeleton recognition, and the surrounding environment detection unit 18 is used. Using the human position information acquired from the detection result, the orientation of the human face portion with respect to the robot 10 is detected.

なお、向き検出部１９としては、同様の検出が可能な他のセンサや装置類を利用しても良い。また、周囲環境検出部１８及び向き検出部１９は、前述の各種情報を取得できる限りにおいて、一体化された装置やシステムにより構成することもできる。 As the orientation detection unit 19, other sensors and devices capable of similar detection may be used. Further, the ambient environment detection unit 18 and the orientation detection unit 19 can be configured by an integrated device or system as long as the above-mentioned various information can be acquired.

前記制御装置１３では、予め設定されたスタート地点から所定のゴール地点まで、障害物への影響を極力低減しながらロボット１０が自律移動可能となるように、最適となる移動経路を探索し、動作部１１への動作指令が行われる。 The control device 13 searches for and operates an optimum movement route so that the robot 10 can autonomously move from a preset start point to a predetermined goal point while reducing the influence on obstacles as much as possible. An operation command is given to the unit 11.

この制御装置１３は、ロボット１０に一体的に或いは別体として設けられており、ＣＰＵ等の演算処理装置及びメモリやハードディスク等の記憶装置等からなるコンピュータによって構成されている。当該コンピュータには、以下の各手段として機能させるためのプログラムがインストールされている。 The control device 13 is provided integrally with the robot 10 or as a separate body, and is composed of a computer including an arithmetic processing unit such as a CPU and a storage device such as a memory and a hard disk. A program for functioning as each of the following means is installed in the computer.

次に、前記制御装置１３の具体的構成について説明する。 Next, a specific configuration of the control device 13 will be described.

前記制御装置１３は、周囲環境検出部１８での検出結果を用いて、将来的に人間がロボット１０に干渉する可能性を推定する干渉予測手段２１と、向き検出部１９での検出結果を用いて、当該検出時点でのロボット１０に対する人間の認知状況を推定する認知状況推定手段２２と、干渉予測手段２１及び認知状況推定手段２２での推定結果に応じて所定の情報を記憶する記憶手段２３と、ロボット１０がゴール地点に向かって移動する際に、障害物の状況に応じて、ロボット１０の移動経路の候補となる候補経路を複数生成する候補経路探索手段２４と、これら各候補経路の中から最適となる最適経路を抽出する最適経路抽出手段２５と、最適経路に沿ったロボット１０の自律移動が可能となるように、動作部１１に動作指令する動作指令手段２６とを備えている。 The control device 13 uses the interference prediction means 21 for estimating the possibility that a human interferes with the robot 10 in the future by using the detection result in the ambient environment detection unit 18, and the detection result in the orientation detection unit 19. The cognitive situation estimating means 22 for estimating the human cognitive situation for the robot 10 at the time of the detection, and the storage means 23 for storing predetermined information according to the estimation results of the interference predicting means 21 and the cognitive situation estimating means 22. The candidate route search means 24 that generates a plurality of candidate routes that are candidates for the movement route of the robot 10 according to the situation of obstacles when the robot 10 moves toward the goal point, and each of these candidate routes. It is provided with an optimum route extraction means 25 for extracting the optimum optimum route from the inside, and an operation command means 26 for instructing the operation unit 11 to operate so that the robot 10 can autonomously move along the optimum route. ..

前記干渉予測手段２１では、本発明者らが既に提案した特開２０２０−４６７５９号公報等に開示されているように、周囲環境検出部１８での検出結果から人間の現在の移動状況を取得し、ロボット１０が将来的に人間に干渉する可能性の有無が判定される。すなわち、ここでは、ロボット１０と人間の周囲に所定のサイズのパーソナルエリアが設定され、将来的に、これらパーソナルエリア同士が少なくとも一部でも重なり合うと予測される場合には、干渉の可能性が有ると判定され、そうでない場合には、干渉の可能性が無いと判定される。具体的には、周囲環境検出部１８での検出結果による人間の速度ベクトルに基づいて、ロボット１０と人間が横並びになった位置から、ロボット１０と人間の各体幹中心の間の距離を計算する。そして、当該距離が各パーソナルエリアの離間距離よりも短いときに、干渉の可能性が有る「干渉度有」と判定され、そうでないときに、干渉の可能性が無い「干渉可能性無」と判定される。 In the interference prediction means 21, as disclosed in Japanese Patent Application Laid-Open No. 2020-46759 already proposed by the present inventors, the current movement state of a human being is acquired from the detection result by the ambient environment detection unit 18. , Whether or not the robot 10 may interfere with humans in the future is determined. That is, here, if a personal area of a predetermined size is set around the robot 10 and a human being, and it is predicted that at least a part of these personal areas will overlap each other in the future, there is a possibility of interference. If not, it is determined that there is no possibility of interference. Specifically, the distance between the robot 10 and the center of each human trunk is calculated from the position where the robot 10 and the human are laid side by side based on the speed vector of the human based on the detection result by the surrounding environment detection unit 18. do. Then, when the distance is shorter than the separation distance of each personal area, it is determined that there is a possibility of interference, "there is a degree of interference", and when it is not, there is no possibility of interference, "there is no possibility of interference". It is judged.

前記認知状況推定手段２２では、向き検出部１９で検知した人間の顔の向きに基づき、次のようにして、ロボット１０に対する人間の認知状況を推定するようになっている。すなわち、先ず、周囲の人間や物体を認知可能となる視野範囲が認知可能範囲として特定される。ここで、人間の認知可能範囲は、人間の視野角（例えば、２００度）の範囲として予め設定される。なお、ロボット１０の認知可能範囲は、検出装置１２での障害物の検知範囲とされる。そして、ロボット１０と人間の相対距離が予め設定された距離以下になったときに、ロボット１０を視認する目が存在する人間の顔の向きから、人間の認知可能範囲内にロボット１０が存在する場合には、人間がロボット１０を認知している「認知度有」と判定される。一方、そうでない場合には、人間がロボット１０を認知していない「認知度無」と判定される。 The cognitive status estimation means 22 estimates the human cognitive status for the robot 10 as follows, based on the orientation of the human face detected by the orientation detecting unit 19. That is, first, the visual field range in which the surrounding human beings and objects can be recognized is specified as the cognitive range. Here, the human recognizable range is preset as a range of the human viewing angle (for example, 200 degrees). The perceptible range of the robot 10 is the detection range of obstacles by the detection device 12. Then, when the relative distance between the robot 10 and the human becomes equal to or less than a preset distance, the robot 10 exists within the human recognizable range from the direction of the human face where the eyes for visually recognizing the robot 10 exist. In that case, it is determined that the human being recognizes the robot 10 as "recognitional". On the other hand, if this is not the case, it is determined that the human does not recognize the robot 10 as "no recognition".

前記記憶手段２３では、ロボット１０の周囲の各人間について、干渉予測手段２１により干渉度有と判定されたときに、当該人間の位置情報や速度情報と、認知状況推定手段２２で推定された当該人間の認知状況とが記憶される。 In the storage means 23, when each human around the robot 10 is determined to have an interference degree by the interference predicting means 21, the position information and speed information of the human and the cognitive situation estimating means 22 are estimated. The human cognitive status is memorized.

前記候補経路探索手段２４では、本発明者らが既に提案した特開２０２０−４６７５９号公報等に開示された手法等を用い、ロボット１０の候補経路を複数生成するようになっている。すなわち、ここでは、図２に示されるように、現在のロボット１０の位置をスタート地点ＳＰとし、当該スタート地点ＳＰとゴール地点ＧＰとを直線で結んだ直線経路（同図中破線）が設定される。そして、当該直線経路において、干渉予測手段２１での判定により人間Ｈとの間での干渉が将来的に生じ得る場合、人間Ｈの左右両側となる横に、人間Ｈとの干渉を回避する経路の通過点（ＷａｙＰｏｉｎｔ：以下、「ＷＰ」と称する）が設定される。更に、当該ＷＰとゴール地点ＧＰとを直線で結んだ別の直線経路上において、他の人間Ｈとの間での干渉が将来的に生じ得る場合に、当該他の人間Ｈの左右両側となる横に同様にして次のＷＰが設定され、この処理が繰り返し行われる。最後に、このような手順で設定された各ＷＰを順に通過する候補経路ＰＴ（図２の例では４通り）が生成される。 The candidate route search means 24 uses the method disclosed in Japanese Patent Application Laid-Open No. 2020-46759 already proposed by the present inventors to generate a plurality of candidate routes for the robot 10. That is, here, as shown in FIG. 2, the current position of the robot 10 is set as the start point SP, and a straight line path (broken line in the figure) connecting the start point SP and the goal point GP with a straight line is set. To. Then, in the linear path, if interference with the human H may occur in the future due to the determination by the interference predicting means 21, a path for avoiding the interference with the human H on the left and right sides of the human H. (Way Point: hereinafter referred to as "WP") is set. Further, on another straight line path connecting the WP and the goal point GP with a straight line, if interference with another human H may occur in the future, it will be on the left and right sides of the other human H. The next WP is set sideways in the same manner, and this process is repeated. Finally, candidate route PTs (4 ways in the example of FIG. 2) that sequentially pass through each WP set by such a procedure are generated.

以上のように複数生成される候補経路としては、ロボット１０が人間Ｈのパーソナルエリアに侵入しない回避経路と、ロボット１０が人間Ｈのパーソナルエリアに侵入するが、ロボット１０と人間Ｈ同士の体幹が接触しない接近経路と、ロボット１０のアーム１５を使って人間に接触しながら人間を避ける接触経路とがある。本実施形態において、当該接触経路は、後述する移動予測シミュレーション部２８でのシミュレーション結果により、後述するデッドロック状態がロボット１０に発生し得る場合にのみ生成されるが、当該条件に限定されるものではない。 As a plurality of candidate routes generated as described above, an avoidance route in which the robot 10 does not invade the personal area of human H and a trunk in which the robot 10 invades the personal area of human H, but the robot 10 and the human H are connected to each other. There is an approach path that the robot 10 does not contact, and a contact path that avoids the human while contacting the human using the arm 15 of the robot 10. In the present embodiment, the contact path is generated only when the deadlock state described later can occur in the robot 10 based on the simulation result in the movement prediction simulation unit 28 described later, but is limited to the conditions. is not it.

前記最適軌道抽出手段２５は、候補経路探索手段２４で探索された各候補経路について、その経路に沿ってロボット１０が移動するときに、その周囲に存在する人間とロボット１０の相互作用によるこれらの移動を経時的にシミュレーションする移動予測シミュレーション部２８と、移動予測シミュレーション部２８でのシミュレーション結果から、ロボット１０の移動効率及び周囲の人間に与える影響を考慮して各候補経路の中から最適経路を決定する最終決定部２５とを備えている。 The optimum orbit extraction means 25 is for each candidate route searched by the candidate route search means 24, when the robot 10 moves along the route, these are due to the interaction between the human and the robot 10 existing around the robot 10. From the simulation results of the movement prediction simulation unit 28 that simulates the movement over time and the movement prediction simulation unit 28, the optimum route is selected from each candidate route in consideration of the movement efficiency of the robot 10 and the influence on surrounding humans. It is provided with a final determination unit 25 for determining.

前記移動予測シミュレーション部２８では、運動方程式に基づく移動予測モデルである従来のＳＦＭ（ＳｏｃｉａｌＦｏｒｃｅＭｏｄｅｌ）をベースにして本発明者らが拡張した新たな移動予測モデルにより、ゴール地点にロボット１０が移動するまでの間において、ロボット１０の移動とその周囲の障害物の相互作用を考慮したロボット１０及び人間の経時的な移動状態をシミュレーションするようになっている。 In the movement prediction simulation unit 28, the robot 10 moves to the goal point by a new movement prediction model extended by the present inventors based on the conventional SFM (Social Force Model) which is a movement prediction model based on the equation of motion. In the meantime, the movement state of the robot 10 and the human being with time is simulated in consideration of the movement of the robot 10 and the interaction of obstacles around the robot 10.

前記ＳＦＭは、人間やロボット等の移動可能な各エージェントに作用する仮想的な力ベクトルの方向に各エージェントが移動するとして、当該各エージェントの移動を予測するモデルである。このＳＦＭでは、対象エージェントにおける目的地からの仮想的な引力と、壁等の固定障害物のように静止環境下にあるオブジェクトや他のエージェントからの仮想的な斥力とを合成した合力ベクトルが求められ、当該合力ベクトルに従って対象エージェントが移動すると推定される。ところが、ロボット１０の移動に伴う周囲の人間に与える物理的負担や心理的負担を考慮して最適経路を生成するに際し、従来のＳＦＭを単に利用してロボット１０の周囲の人間の移動予測を行うと、次の理由により、最適となる移動経路の抽出を必ずしも十分に行えない。先ず、第１の理由として、従来のＳＦＭでは、対象となるエージェントの後方から接近する他の人間等のエージェントからの斥力も考慮されてしまうが、実空間においては、人間が後方の人間を認知して避ける行動は不自然である。また、第２の理由として、従来のＳＦＭでは、ロボット１０が移動している際に、他のエージェントの状態や目的地の位置により、前記斥力と前記引力が釣り合ってしまうデッドロック状態が生じてしまう。このような場合に、ロボット１０が最も斥力を受ける他のエージェントに対して接触を用いた働きかけを行う接触経路が候補経路として生成されるが、この際の接触力は従来のＳＦＭで考慮されていない。 The SFM is a model that predicts the movement of each agent assuming that each agent moves in the direction of a virtual force vector acting on each movable agent such as a human or a robot. In this SFM, a resultant force vector that combines the virtual attractive force from the destination of the target agent and the virtual repulsive force from an object in a stationary environment such as a fixed obstacle such as a wall or another agent is obtained. It is estimated that the target agent moves according to the resultant force vector. However, when generating the optimum route in consideration of the physical burden and the psychological burden on the surrounding humans due to the movement of the robot 10, the conventional SFM is simply used to predict the movement of the humans around the robot 10. And, for the following reasons, it is not always possible to sufficiently extract the optimum movement route. First, as the first reason, in the conventional SFM, the repulsive force from agents such as other humans approaching from behind the target agent is also taken into consideration, but in the real space, the human recognizes the person behind. The behavior of avoiding it is unnatural. Further, as a second reason, in the conventional SFM, when the robot 10 is moving, a deadlock state occurs in which the repulsive force and the attractive force are balanced depending on the state of another agent or the position of the destination. It ends up. In such a case, a contact path is generated as a candidate path in which the robot 10 acts on another agent that receives the most repulsive force by using contact, but the contact force at this time is considered in the conventional SFM. No.

そこで、本発明者らが新たに創出した拡張モデルでは、認知状況推定手段２２での認知度と、候補経路探索手段２４で求めた接触経路において対象の人間に作用する接触力とを更に考慮し、ロボット１０の移動過程において、各エージェント、すなわち、ロボット１０及びその周囲に存在する人間の移動予測がなされる。 Therefore, in the extended model newly created by the present inventors, the degree of recognition in the cognitive situation estimation means 22 and the contact force acting on the target human in the contact path obtained by the candidate route search means 24 are further considered. In the process of moving the robot 10, the movement of each agent, that is, the robot 10 and the human beings around the robot 10 is predicted.

ここでの移動予測シミュレーションは、各候補経路について、それぞれ設定されたＷＰ間で時系列に沿ってそれぞれ順に行われる。つまり、スタート地点から最初のＷＰをロボット１０の目的地とした区間で最初のシミュレーションが行われた後、その状態から次のＷＰにロボット１０の目的地を変更した次の区間で同様のシミュレーションが行われ、当該シミュレーションがロボット１０のゴール地点までのＷＰ間の各区間において経時的に行われる。 The movement prediction simulation here is performed in order for each candidate route in chronological order between the set WPs. That is, after the first simulation is performed in the section where the first WP from the start point is the destination of the robot 10, the same simulation is performed in the next section where the destination of the robot 10 is changed from that state to the next WP. The simulation is performed, and the simulation is performed over time in each section between the WPs to the goal point of the robot 10.

具体的に、ここでは、各区間でのシミュレーション毎に、各エージェントの移動予測を行うに際し、予測対象となる対象エージェント毎に、それぞれ作用する合力ベクトルＦが計算される。当該合力ベクトルＦは、ロボット１０の仮想移動の際の所定のタイミング毎（例えば、０．５秒毎）に計算され、当該タイミング毎において、各エージェントが対応する合力ベクトルＦに従って移動するものと予測される。なお、シミュレーション上、各人間に設定される目的地については、検出装置１２での検出結果に基づき算出された移動方向に沿って仮想的に設定された所定地点とされる。 Specifically, here, the resultant force vector F that acts on each target agent to be predicted is calculated for each target agent to be predicted when predicting the movement of each agent for each simulation in each section. The resultant force vector F is calculated at predetermined timings (for example, every 0.5 seconds) at the time of virtual movement of the robot 10, and it is predicted that each agent moves according to the corresponding resultant force vector F at each timing. Will be done. In the simulation, the destination set for each person is a predetermined point virtually set along the moving direction calculated based on the detection result by the detection device 12.

前記合力ベクトルＦは、対象エージェントにおける目的地からの引力ベクトルＦ^Ｇと、対象エージェントの周囲に存在する他のエージェントｉからの斥力ベクトルＦ_ｉ ^Ｈの合計と、対象エージェントの周囲に存在する壁等の静止障害物であるオブジェクトｊからの斥力ベクトルＦ_ｊ ^ｏの合計と、接触経路においてロボット１０が人間に付加する接触力ベクトルＦ_ｋ ^Ｃとの合計となり、次式によって算出される。

The resultant force vector F is a attractive force vector F ^G from the destination in the subject agents, the sum of the repulsive force vector F _i ^H from other agents i present around the target agent, a wall or the like present around the target agent sum of the repulsive force vector F _j ^o from the object j is a stationary obstacle, the robot 10 is the sum of the contact force vector F _k ^C to be added to the human in the contact path is calculated using the following formula.

上式（２）において、ｍは、対象エージェントの重量であり、ｖ^０（ｔ）は、対象エージェントの希望速度ベクトルであり、これらは、ロボット１０や人間等のエージェントの種類に応じて予め所定値が設定される。また、ｖ（ｔ）は、対象エージェントの現在速度ベクトルであり、ロボット１０であれば把握されており、人間であれば検出装置１２での検出結果から推定される。更に、τは、シミュレーションを行う前記タイミングの間隔に相当する予め設定された時定数である。 In the above equation (2), m is the weight of the target agent, v ⁰ (t) is the desired speed vector of the target agent, and these are predetermined according to the type of the agent such as the robot 10 or a human. The value is set. Further, v (t) is the current velocity vector of the target agent, which is grasped by the robot 10 and estimated from the detection result by the detection device 12 if it is a human. Further, τ is a preset time constant corresponding to the timing interval at which the simulation is performed.

上式（３）、（４）において、ｆｐ_ｉ ^Ｈ、ｆｐ_ｊ ^Ｏは、他のエージェントｉ、オブジェクトｊへの衝突時に発生する力ベクトルと定義され、ｆｓ_ｉ ^Ｈ、ｆｓ_ｊ ^Ｏは、他のエージェントｉ、オブジェクトｊに対する心理的要素に影響する心理的な斥力ベクトルと定義され、次式で算出される。

ここで、ｋｐ、ｋｓは、斥力の大きさに関する係数であり、事前になされた実験等の結果に基づいて、エージェントの動きが自然になる所定の一定値に設定される。また、ｄ_ｉ、ｄ_ｊは、斥力を発する他のエージェントｉやオブジェクトｊと対象エージェントとの離間距離であり、ｖ_ｉ、ｖ_ｊは、これらの間の単位方向速度ベクトルであり、それぞれ、検出装置１２での検出結果から特定される。更に、ｒは、ロボット１０や人間のすれ違いに必要となる円形の最小領域の半径であり、ｓは、対象エージェントに斥力が及ぼされる範囲に相当するパーソナルエリアの半径である。これらｒ、ｓは、例えば、ｒ＝０．６ｍ、ｓ＝１ｍ等の一定値に予め設定される。更に、θは、従来のＳＦＭにおけるランプ関数である。 In the above equations (3) and (4), fp _i ^H and fp _j ^O are defined as force vectors generated when colliding with another agent i and object j, and fs _i ^H and fs _j ^O are other. It is defined as a psychological repulsive force vector that affects the psychological elements of the agent i and the object j, and is calculated by the following equation.

Here, kp and ks are coefficients related to the magnitude of the repulsive force, and are set to predetermined constant values that make the movement of the agent natural based on the results of experiments and the like performed in advance. Also, d _i, d _j is the distance between the other agents i and object j and the target agent that emits repulsive force, v _i, v _j is a unit direction velocity vector between them, respectively, detected It is specified from the detection result in the device 12. Further, r is the radius of the minimum circular region required for the robot 10 and humans to pass each other, and s is the radius of the personal area corresponding to the range in which the repulsive force is applied to the target agent. These r and s are preset to constant values such as r = 0.6m and s = 1m. Further, θ is a ramp function in the conventional SFM.

また、上式（３）、（４）での各斥力ベクトルＦ_ｉ ^Ｈ、Ｆ_ｊ ^ｏの算出においては、認知状況推定手段２２での認知度が考慮される。すなわち、上式（５）の通り、対象エージェントが認知していない他のエージェントやオブジェクトから心理的な斥力を受けないように、認知度ａｗがゼロ、すなわち、「認知度無」の場合には、心理的な斥力ベクトルｆｓ_ｉ ^Ｈ、ｆｓ_ｊ ^Ｏをゼロとして、各斥力ベクトルＦ_ｉ ^Ｈ、Ｆ_ｊ ^ｏが求められる。換言すれば、「認知度有」の場合にのみ心理的な斥力ベクトルｆｓ_ｉ ^Ｈ、ｆｓ_ｊ ^Ｏが加味され、作用することになる。 Further, the above equation (3), each of the repulsive force vector _F ⁱ H in _(4), in the calculation of the _F ^{j o,} awareness of cognitive state estimation means 22 is considered. That is, as shown in the above equation (5), when the recognition level a is zero, that is, "no recognition level" so as not to receive psychological repulsion from other agents or objects that the target agent does not recognize. psychological repulsion vector _fs ⁱ H, a _fs ^{j O} as zero, the repulsive force vector _{_F} ⁱ _H, _F ^{j o} is obtained. In other words, the psychological repulsive force vectors fs _i ^H and fs _j ^O are added and acted only in the case of "recognition".

以上の移動予測シミュレーション部２８では、例えば、図３に示されるように、ロボット１０及び周囲の人間Ａ〜Ｄの移動予測シミュレーションが行われる。つまり、ロボット１０及び周囲の人間Ａ〜Ｄの各エージェントについて、認知状況推定手段２２で特定されたそれぞれの認知可能範囲Ｒに応じて、一定タイミング毎に合力ベクトルＦがそれぞれ算出される。そして、当該合力ベクトルＦから、各エージェントの移動状態が推定される。 In the above-mentioned movement prediction simulation unit 28, for example, as shown in FIG. 3, the movement prediction simulation of the robot 10 and the surrounding humans A to D is performed. That is, for each of the agents of the robot 10 and the surrounding humans A to D, the resultant force vector F is calculated at regular timings according to the cognitive range R specified by the cognitive status estimation means 22. Then, the moving state of each agent is estimated from the resultant force vector F.

先ず、候補経路内に設定されたＷＰがロボット１０の目的地Ｐとして設定され、その後、各エージェントの移動予測シミュレーションが行われる。図３（Ａ）のタイミングでは、ロボット１０が壁Ｗから斥力（同図中破線矢印）を受け、その影響により、目的地Ｐとの直線経路よりもやや内側方向にロボット１０が移動する。そして、ロボット１０が目的地Ｐに向かって更に進んだ同図（Ｂ）のタイミングでのロボット１０の移動は、人間Ｂからの斥力（同図中一点鎖線矢印）にも影響を受ける。その後、同図（Ｃ）のようにロボット１０が更に進むと、同図（Ｄ）に示されるように、ロボット１０に接近状態となる人間Ａが、ロボット１０からの接触力（同図中太線矢印）により同図中右側に移動し、人間Ａの移動により、人間Ｂ，Ｃに斥力（同図中一点鎖線矢印）が伝播して、同図（Ｅ）に示されるように、人間Ｂ，Ｃも移動することになる。そして、同図（Ｆ）に示されるように、ロボット１０が目的地Ｐに達したら、次のＷＰが目的地として設定され、同様にして移動予測シミュレーションが行われる。 First, the WP set in the candidate route is set as the destination P of the robot 10, and then the movement prediction simulation of each agent is performed. At the timing of FIG. 3A, the robot 10 receives a repulsive force (broken line arrow in the figure) from the wall W, and due to the influence, the robot 10 moves slightly inward from the straight path to the destination P. Then, the movement of the robot 10 at the timing of the figure (B) in which the robot 10 further advances toward the destination P is also affected by the repulsive force from the human B (one-dot chain arrow in the figure). After that, when the robot 10 advances further as shown in the figure (C), as shown in the figure (D), the human A who is in a state of approaching the robot 10 has a contact force from the robot 10 (thick line in the figure). The arrow) moves to the right side in the figure, and the movement of the human A causes the repulsive force (one-dot chain arrow in the figure) to propagate to the humans B and C, and as shown in the figure (E), the human B, C will also move. Then, as shown in the figure (F), when the robot 10 reaches the destination P, the next WP is set as the destination, and the movement prediction simulation is performed in the same manner.

前記最終決定部２９では、各候補経路について、予め記憶された数式により、ロボット１０及びその周囲の人間の移動負荷に対応する次の２種の指標値が算出され、これら指標値を総合した経路コストにより、各候補経路の中から最も適切となる最適経路が選択される。 In the final determination unit 29, the following two types of index values corresponding to the movement load of the robot 10 and the humans around it are calculated by the mathematical formula stored in advance for each candidate route, and the route in which these index values are integrated is calculated. Depending on the cost, the most appropriate optimal route is selected from each candidate route.

当該最終決定部２９での具体的な処理内容を以下に説明する。 The specific processing contents in the final determination unit 29 will be described below.

先ず、前記指標値として、ロボット１０の移動効率に関する第１の指標値と、移動予測シミュレーション部２８での移動予測結果に基づくロボット１０の周囲に存在する人間に与える負荷に関する第２の指標値とが求められる。 First, as the index values, a first index value relating to the movement efficiency of the robot 10 and a second index value relating to a load applied to a human being around the robot 10 based on the movement prediction result of the movement prediction simulation unit 28. Is required.

前記第１の指標値は、移動シミュレーションにおける目的地に対するロボット１０の進行方向に係る第１のパラメータと、当該進行方向に直交し、ロボット１０が人間や物体を回避するためのロボット１０の回避方向に係る第２のパラメータとを合計した移動コストＥに対応する。 The first index value is orthogonal to the first parameter related to the traveling direction of the robot 10 with respect to the destination in the movement simulation, and is orthogonal to the traveling direction, and the avoidance direction of the robot 10 for the robot 10 to avoid humans and objects. Corresponds to the movement cost E, which is the sum of the second parameter according to the above.

前記第１のパラメータでは、ロボット１０が一定速度で直進し続ける行動を最もコストの低い行動とし、ロボット１０の進行方向への加減速度をコストの増減パラメータとしている。 In the first parameter, the action in which the robot 10 continues to travel straight at a constant speed is the lowest cost action, and the acceleration / deceleration in the traveling direction of the robot 10 is the cost increase / decrease parameter.

前記第２のパラメータでは、ロボット１０が干渉回避を行う際の回避幅が大きくなる程、ロボット１０の移動効率が悪くなることから、ロボット１０の進行方向に直交する回避方向の移動をコストの増減パラメータとしている。 In the second parameter, the larger the avoidance width when the robot 10 avoids interference, the worse the movement efficiency of the robot 10. Therefore, the cost of moving in the avoidance direction orthogonal to the traveling direction of the robot 10 is increased or decreased. It is a parameter.

つまり、前記移動コストＥは、スタート地点からゴール地点までの間において、移動予測シミュレーション部２８でシミュレーションされた各ＷＰ間でのロボット１０の移動状態の変化を総合して、次式により算出される。

上式において、ｖ_ｙｔは、時刻ｔにおけるロボット１０の進行方向の速度であり、ｖ_ｙ０は、同進行方向の基準速度であり、ｖ_ｘｔは、同回避方向の速度であり、ｖ_ｘｔ−１は、１ステップ前の同回避方向の速度である。また、Ａは、前記進行方向の重み係数であり、Ｂは、前記回避方向の重み係数であり、所定値（例えば、Ａ＝１、Ｂ＝１）が予め設定される。 That is, the movement cost E is calculated by the following equation by integrating the changes in the movement state of the robot 10 between each WP simulated by the movement prediction simulation unit 28 from the start point to the goal point. ..

In the above equation, v _yt is the speed in the traveling direction of the robot 10 at time t, v _y0 is the reference speed in the same traveling direction, v _xt is the speed in the same avoiding direction, and v _xt-1. Is the speed in the same avoidance direction one step before. Further, A is a weighting coefficient in the traveling direction, B is a weighting coefficient in the avoiding direction, and predetermined values (for example, A = 1, B = 1) are set in advance.

前記第２の指標値は、スタート地点からゴール地点までのロボット１０の移動時に、周囲の人間の移動及び心理的な影響をより与えない経路を選択する必要があることから、移動予測シミュレーション部２８でのシミュレーション結果により求められた各エージェントの斥力ベクトルＦ_ｉ ^Ｈ、Ｆ_ｊ ^ｏの総和と、ロボット１０が人間に接触する際の接触力ベクトルＦ_ｋ ^Ｃを総合した影響度Ｄである。この影響度Ｄは、次式により求められる。

上式において、ｎはロボット１０が認識した人数に相当する。 As for the second index value, when the robot 10 moves from the start point to the goal point, it is necessary to select a route that does not have a greater influence on the movement and psychological effects of surrounding humans. Therefore, the movement prediction simulation unit 28 is a simulation repulsive force vector of each agent obtained as a result F _{i ^H,} F _j ^o and the sum of the degree of influence D obtained by integrating the contact force vector F _k ^C when the robot 10 is in contact with the human at. This degree of influence D is obtained by the following equation.

In the above equation, n corresponds to the number of people recognized by the robot 10.

次に、次式のように、移動コストＥ及び影響度Ｄに重み係数ａ、ｂを乗じた上で加算することにより、各候補経路について経路コストＣがそれぞれ求められる。なお、重み係数ａ、ｂは、事前の実験結果により一定値に設定され、例えば、ａ＝１．７、ｂ＝０．３としている。

Next, the route cost C is obtained for each candidate route by multiplying the movement cost E and the degree of influence D by the weight coefficients a and b and then adding them as in the following equation. The weighting coefficients a and b are set to constant values according to the results of prior experiments, and for example, a = 1.7 and b = 0.3.

そして、経路コストＣの最も小さい候補経路が最適経路として決定され、この最適経路に沿ったロボット１０の移動経路が生成されることになる。ここで、当該最適経路における各ＷＰ間のロボット１０の移動は、前記シミュレーション結果による移動軌跡に沿って行われる。 Then, the candidate route having the smallest route cost C is determined as the optimum route, and the movement route of the robot 10 along this optimum route is generated. Here, the movement of the robot 10 between the WPs in the optimum route is performed along the movement locus according to the simulation result.

次に、本実施形態の変形例について説明する。なお、以下の説明において、前記実施形態と同一若しくは同等の構成部分については同一符号を用いるものとし、前記実施形態と相違する事項を除き、説明を省略若しくは簡略にする。 Next, a modification of the present embodiment will be described. In the following description, the same reference numerals will be used for the same or equivalent components as those in the above embodiment, and the description will be omitted or simplified except for matters different from those in the above embodiment.

本変形例では、制御装置１３により、移動障害物のうちロボット１０の周囲から接近する人間（接近者）を対象エージェントとし、当該接近者が将来的に移動し得る複数の経路を候補経路として予測し、その中で最適となる最適経路を特定し、当該最適経路での接近者の移動に合わせて接近者を避けることができるように、ロボット１０の移動経路を決定するようになっている。 In this modification, the control device 13 predicts a human (approacher) approaching from around the robot 10 among moving obstacles as a target agent, and predicts a plurality of routes that the approacher can move in the future as candidate routes. Then, the optimum route that becomes the optimum is specified, and the movement route of the robot 10 is determined so that the approaching person can be avoided according to the movement of the approaching person on the optimum route.

すなわち、本変形例に係る候補経路探索手段２４では、前記実施形態においてロボット１０に対して行ったのと同様の手順により、ロボット１０や他の人間等の他のエージェントを避けて通る接近者の将来的な経路を候補経路として予測するようになっており、当該候補経路は、接近者が他のエージェントに接触する前述の接触経路を含め、複数通り導出される。 That is, in the candidate route search means 24 according to the present modification, the approaching person who passes by avoiding other agents such as the robot 10 and other humans by the same procedure as that performed for the robot 10 in the above embodiment. A future route is predicted as a candidate route, and the candidate route is derived in a plurality of ways including the above-mentioned contact route in which an approaching person contacts another agent.

また、本変形例に係る最適経路抽出手段２５では、先ず、移動予測シミュレーション部２８において、前記実施形態と同様の手順により、接近者の各候補経路について、当該接近者の移動に伴うロボット１０の移動及び各障害物の相互作用により、各エージェントの経時的な移動状態がシミュレーションされる。そして、最終決定部２９では、前記実施形態に対しロボット１０を対象エージェントとなる接近者に代えた接近者の移動効率に関する第１の指標値である移動コストＥと、前記実施形態と同様の人間に与える負荷に関する第２の指標値である影響度Ｄとが求められる。その後は、前記実施形態と同様にして、対象となる接近者の移動コストＣが求められ、当該移動コストＣが最小となる候補経路が当該接近者の最適経路となる。 Further, in the optimum route extraction means 25 according to the present modification, first, in the movement prediction simulation unit 28, the robot 10 accompanies the movement of the approaching person for each candidate route of the approaching person by the same procedure as that of the embodiment. The movement and the interaction of each obstacle simulate the movement state of each agent over time. Then, in the final determination unit 29, the movement cost E, which is the first index value regarding the movement efficiency of the approaching person who replaced the robot 10 with the approaching person as the target agent for the embodiment, and the human being similar to the embodiment. The degree of influence D, which is the second index value regarding the load applied to the robot, is obtained. After that, the movement cost C of the target approaching person is obtained in the same manner as in the above embodiment, and the candidate route having the minimum movement cost C becomes the optimum route of the approaching person.

そして、最終決定部２９では、移動予測シミュレーション部２８によるシミュレーション結果により、対象の接近者が最適経路を移動するときのロボット１０の経路がロボット１０の移動経路として特定される。すなわち、当該移動経路は、対象の接近者が最適経路を移動する条件でシミュレーションされたロボット１０の経路である。更に、動作指令手段２６では、ロボット１０が決定された移動経路に沿って自律移動するように、動作部１１への動作指令がなされる。 Then, in the final determination unit 29, the path of the robot 10 when the target approaching person moves on the optimum path is specified as the movement path of the robot 10 by the simulation result by the movement prediction simulation unit 28. That is, the movement route is the route of the robot 10 simulated under the condition that the target approacher moves on the optimum route. Further, the operation command means 26 issues an operation command to the operation unit 11 so that the robot 10 autonomously moves along the determined movement path.

以上の変形例によれば、ロボット１０が人混みを避けて目標地点に移動する前記実施形態での動作制御態様に加え、又は、当該制御態様と選択的に、既に人混みにいるロボット１０が、その横を通り抜けようとする接近者を除けるように動作させる制御態様が可能になる。 According to the above modification, in addition to the motion control mode in the embodiment in which the robot 10 moves to the target point while avoiding the crowd, or selectively with the control mode, the robot 10 already in the crowd is the robot 10. It enables a control mode that operates so as to exclude an approaching person who is trying to pass by.

以上で説明した制御装置１３には、エージェントにおける周囲の認知状況を推定する認知状態推定手段２２と、エージェントの経時的な移動をシミュレーションする移動予測シミュレーション部２８（移動予測シミュレーション手段）とが含まれている。従って、制御装置１３は、所定のエージェントが移動する過程で、その周囲に移動可能な他のエージェントと、常時静止状態となるオブジェクトとが存在する際に、各エージェントの経時的な移動を予測する移動予測装置としても機能する。 The control device 13 described above includes a cognitive state estimation means 22 that estimates the cognitive situation of the surroundings of the agent, and a movement prediction simulation unit 28 (movement prediction simulation means) that simulates the movement of the agent over time. ing. Therefore, the control device 13 predicts the movement of each agent over time when there are other agents that can move around the predetermined agent and an object that is always in a stationary state in the process of moving the predetermined agent. It also functions as a movement prediction device.

また、制御装置１３は、前記候補経路探索手段２４と前記最適経路抽出手段２５とを有しており、周囲に存在する障害物を避けながら自律移動する自律移動するロボット１０の移動経路を生成する移動経路生成装置としても機能する。 Further, the control device 13 has the candidate route search means 24 and the optimum route extraction means 25, and generates a movement route of the autonomously moving robot 10 that autonomously moves while avoiding obstacles existing in the surroundings. It also functions as a movement route generator.

なお、本発明に係るロボットとしては、前記実施形態で説明した自律移動型のロボット１０に限定されるものではなく、自動車両、船舶、飛行体等、所定の空間内を自律的に移動可能な移動体の他に、所定範囲の空間内で移動するロボットアーム等のマニピュレータであってもよく、これら移動時における前記最適経路の生成を前述と同様の構成及び手順により行うことができる。 The robot according to the present invention is not limited to the autonomous mobile robot 10 described in the above embodiment, and can autonomously move in a predetermined space such as an automatic vehicle, a ship, or an air vehicle. In addition to the moving body, a manipulator such as a robot arm that moves within a predetermined range of space may be used, and the optimum path can be generated at the time of movement by the same configuration and procedure as described above.

その他、本発明における装置各部の構成は図示構成例に限定されるものではなく、実質的に同様の作用を奏する限りにおいて、種々の変更が可能である。 In addition, the configuration of each part of the device in the present invention is not limited to the illustrated configuration example, and various changes can be made as long as substantially the same operation is achieved.

１０ロボット
１１動作部
１２検出装置
１３制御装置（移動経路生成装置、移動予測装置）
２２認知状態推定手段
２４候補経路探索手段
２５最適経路抽出手段
２６動作指令手段
２８移動予測シミュレーション部（移動予測シミュレーション手段）
２９最終決定部 10 Robot 11 Moving unit 12 Detection device 13 Control device (movement route generator, movement prediction device)
22 Cognitive state estimation means 24 Candidate route search means 25 Optimal route extraction means 26 Motion command means 28 Movement prediction simulation unit (movement prediction simulation means)
29 Final decision department

Claims

In a robot that moves autonomously while avoiding obstacles around it
It is provided with a detection device that detects the position information and speed information of the obstacle, and a control device that controls the autonomous movement while considering the situation of the obstacle from the detection result of the detection device.
The control device is a candidate for a route to which the target agent among the agents consisting of the robot and the movable obstacle among the obstacles moves based on the position information and the speed information of the obstacle. Candidate route search means for generating a plurality of candidate routes, an optimum route extraction means for extracting the optimum optimum route from each of the candidate routes, and an operation command for issuing an operation command for the autonomous movement based on the optimum route. Equipped with means,
The optimum route extraction means is based on a movement prediction simulation unit that simulates the movement state of each agent by the interaction between the robot and the obstacle for each candidate route, and simulation results in the movement prediction simulation unit. It is provided with a final determination unit that determines the optimum route based on the movement efficiency of the target agent and the influence on the movement obstacle.
In the movement prediction simulation unit, the position and speed of each agent are used by utilizing a virtual force that interacts between the agents or between the agents and the object that is an obstacle that is always stationary. Changes are predicted over time,
In the final determination unit, for each of the candidate routes, a first index value regarding the movement efficiency and a second index value regarding the load applied to the movement obstacle are obtained by a mathematical formula stored in advance, and each of these is obtained. A robot characterized in that the candidate route having the lowest route cost obtained by summing up the index values is selected as the optimum route.

In the operation command means, when the robot itself is the target agent, the operation command is given with the optimum path as the movement path of the robot, while the movement obstacle approaching the robot is the target agent. In a certain case, the movement prediction simulation unit issues an operation command to move the robot along the movement path of the robot simulated under the condition that the movement obstacle moves on the optimum path. The robot according to claim 1, which is characterized.

In the candidate route search means, the candidate route including the contact route that avoids the moving obstacle while the target agent is in contact with the other agent is generated.
In the movement prediction simulation unit, a resultant force of a virtual attractive force from the destination of each agent, a virtual repulsive force acting between the object and each agent, and a contact force at the time of contact. The robot according to claim 1, wherein a vector is calculated and it is estimated that each agent moves according to the resultant force vector.

In the final determination unit, as the first index value, the first parameter related to the traveling direction of the target agent and the avoiding direction of the target agent for avoiding the obstacle are orthogonal to the traveling direction. The movement cost, which is the sum of the second parameter, is obtained, and as the second index value, the total influence of the repulsive force and the contact force is obtained, and these movement costs and the influence are multiplied by a weighting coefficient. The robot according to claim 3, wherein the route cost is obtained for each candidate route by adding the above, and the candidate route having the lowest route cost is selected as the optimum route.

The detection device is provided so as to be able to detect the direction of the moving obstacle.
The control device further includes a cognitive state estimation means for estimating the surrounding cognitive state in each agent based on the detection result of the detection device.
In the cognitive state estimation means, for each agent, a visual field range in which humans and objects around the agent can be recognized is specified as a cognitive range.
In the movement prediction simulation unit, as the repulsive force, a force vector generated at the time of a collision and a psychological force vector defined as a psychological element are calculated by a mathematical formula stored in advance, and then the force vector is calculated. Obtained by totaling,
The robot according to claim 3, wherein the psychological force vector is defined to act only when the object or the other agent is within the recognizable range.

In a movement route generator that generates a movement route for a robot that moves autonomously while avoiding obstacles around it.
Based on the position information and speed information of the obstacle, there are a plurality of candidate routes that are candidates for the route to which the target agent among the agents consisting of the robot and the movable obstacle among the obstacles move. It is provided with a candidate route search means to be generated and an optimum route extraction means to extract the optimum optimum route from each of the candidate routes.
The optimum route extraction means is based on a movement prediction simulation unit that simulates the movement state of each agent by the interaction between the robot and the obstacle for each candidate route, and simulation results in the movement prediction simulation unit. It is provided with a final determination unit that determines the optimum route based on the movement efficiency of the target agent and the influence on the movement obstacle.
In the movement prediction simulation unit, the position and speed of each agent are used by utilizing a virtual force that interacts between the agents or between the agents and the object that is an obstacle that is always stationary. Changes are predicted over time,
In the final determination unit, for each of the candidate routes, a first index value regarding the movement efficiency and a second index value regarding the load applied to the movement obstacle are obtained by a mathematical formula stored in advance, and each of these is obtained. When the candidate route having the lowest route cost obtained by summing up the index values is selected as the optimum route and the robot itself is the target agent, the optimum route is used as the movement route, while the robot is used. When the approaching moving obstacle is the target agent, the movement prediction simulation unit sets the path of the robot simulated under the condition that the moving obstacle moves on the optimum path as the moving path. A featured movement route generator.

In the program that makes the computer of the movement route generator function to generate the movement route of the robot that moves autonomously while avoiding the obstacles around it.
Based on the position information and speed information of the obstacle, there are a plurality of candidate routes that are candidates for the route to which the target agent among the agents consisting of the robot and the movable obstacle among the obstacles move. The computer is made to function as a candidate route search means to be generated and an optimum route extraction means for extracting the optimum optimum route from each of the candidate routes.
The optimum route extraction means is based on a movement prediction simulation unit that simulates the movement state of each agent by the interaction between the robot and the obstacle for each candidate route, and simulation results in the movement prediction simulation unit. It is provided with a final determination unit that determines the optimum route based on the movement efficiency of the target agent and the influence on the movement obstacle.
In the movement prediction simulation unit, the position and speed of each agent are used by utilizing a virtual force that interacts between the agents or between the agents and the object that is an obstacle that is always stationary. Changes are predicted over time,
In the final determination unit, for each of the candidate routes, a first index value regarding the movement efficiency and a second index value regarding the load applied to the movement obstacle are obtained by a mathematical formula stored in advance, and each of these is obtained. When the candidate route having the lowest route cost obtained by integrating the index values is selected as the optimum route and the robot itself is the target agent, the optimum route is used as the movement route, while the robot is used. When the approaching moving obstacle is the target agent, the movement prediction simulation unit sets the path of the robot simulated under the condition that the moving obstacle moves on the optimum path as the moving path. A featured travel route generator program.

In a movement prediction device that predicts the movement of each agent over time when there are other agents that can move around the predetermined agent and an object that is always in a stationary state in the process of moving.
By the interaction between the cognitive state estimation means that estimates the surrounding cognitive status of each agent based on the position and orientation of each agent separately acquired, and the agent and the object when the predetermined agent moves, the agent and the object interact with each other. A movement prediction simulation means for simulating the movement state of each agent over time is provided.
In the cognitive state estimation means, for each agent, a visual field range in which humans and objects around the agent can be recognized is specified as a cognitive range.
The movement prediction simulation means uses a mathematical formula stored in advance and separately acquired position information and velocity information of each agent to obtain a virtual attractive force from the destination of each agent and the object or each agent. A resultant force vector of a virtual repulsive force acting between them and a contact force when a predetermined agent comes into contact with another agent is calculated, and it is estimated that each agent moves according to the resultant force vector. As a repulsive force, a force vector generated at the time of a collision and a psychological force vector defined as a psychological element are calculated, and then the force vectors are summed up to obtain the psychological force vector. A movement predictor, characterized in that it is defined to act only when the object or other agent is within the recognizable range.