基於雲端服務之多機械手臂協作運動規劃與控制

Abstract

本文主要探究於多機器手臂在未知環境中的協作任務運動規劃與控制，透過所提出的雲端機器人架構來讓各個機械手臂不僅能達到安全穩定的運作，還能藉由雲端服務完成遠端互動、動作排程與系統監控等功能。當使用者經由手機應用程式察看系統資訊並給出期望的命令時，雲端會使用其資料庫與演算法對該任務進行所有可行動作的安排。排程結果隨後傳遞至機械手臂附近的邊緣節點以進行路徑規劃演算法，倘若無法順利生成完整的無碰撞路徑時，則回傳訊息給雲端做修正或發送通知給使用者。最後，機械手臂執行規劃好的路徑，並經由導納控制完成協作過程中的環境干擾順應，以及用PID控制來最小化系統內力。同時，路徑最佳化演算法會優化尚未被執行的路徑來節省機械手臂的能量損失。從模擬與實驗結果中可驗證出所提之理論與方法的可行性與優良效能。

This thesis mainly addresses the motion planning and control of cooperative tasks with multiple manipulators in an unknown environment. Based on the proposed framework of the cloud robotics system, the manipulators can carry out safe and stable operations, and the cloud can implement functions, such as remote interactions, action scheduling, and system monitoring. When the users view the information of the system and give the commands through mobile applications, the cloud utilizes the database and the algorithms to schedule all feasible actions for the given tasks. After completing action scheduling, the result transmits to the edge nodes near manipulators to execute the path planning algorithm. If there is no collision-free path, the edge nodes will feedback a message to the cloud to do the modifications or to send the notification of the failure to the user. Finally, the planned paths are executed by the manipulators. The admittance control is used to accomplish the compliance for environmental disturbance, and the PID control is used to minimize the internal force during the cooperation. At the same time, the path optimization algorithm optimizes the currently unexecuted paths to further reduce the energy cost of manipulators. According to the simulations and experiments, the proposed theories and methods are verified to have an effective performance.