Control Strategies for Robot Hands

機械手之控制策略

鐘文遠(Wen-Yeuan Chung)

《華岡工程學報》 11期 (1997/06) Pp. 101-117

https://doi.org/10.29987/CCUHKJE.199706.0010

摘要

本計畫發展出兩種適用於機械手系統之整合控制策略。此二策略分別以計算力矩方法及阻抗控制方法為基本架構，並且以三維靈巧操縱為目標。另外，此兩種整合控制策略之發展亦分別考慮以下三種因素並加以整合。
一、各手指間應如何適當分佈施力，以使被抓物不致滑落。二、以力回饋為低層控制迴路以迅速彌補實際施力和所欲施力之差。三、使用經驗累積策略以增進各關節之命令力矩的平順性。
發展出之整合控制策略分別以三維運動的電腦模擬加以測試驗證。電腦模擬中亦考慮到各手指和被抓物體的相對滾動以符合實際情況。對於使用兩種不同控制策略的模擬結果及優缺點加以比較，提出如何選擇適當之控制策略的準則。
所發展出之電腦模擬軟體除可於螢幕上以動畫方式顯示出各機械手指及被抓物體運動外，亦可立即顯示出各機械手指和被抓物體的摩擦角及各手指關節的命令力矩。此外，本軟體除可經些微修改以模擬各種不同的機械手系統。由於此軟體乃用C語言寫成，可順利得和感測器及驅動器等相關硬體介面加以配合，以用於控制實際的機械手系統。

並列摘要

The object of this project is to develop two integrated control strategies for manipulating an object with the use of a multifingered system were developed. Both strategies were developed respectively based on the computed torque method and the impedance control method. The strategies proposed were based on three dimensional manipulation. In terms of developing control strategies, three techniques were considered and combined successfully with each control method. One of the techniques was to find distributed force by optimizing the friction angles to avoid slippage between the grasped object and fingertips. The other was to use force feedback as low level control loop to compensate the difference between actual and desired grasping forces. Another was to use history-based strategy to increase the smoothness of the commanded torques.
Both strategies were tested by running three dimensional simulation examples with the rolling motion between the grasped object and fingertips considered. The simulation results by using different control strategies were compared Further, the way to choose an appropriate control strategy for multifingered systems is also proposed.
The developed simulation software was designed to show animation picture, frition angles, and commanded torques on the monitor. This software can also be used to simulate other multifingered systems with minor modification. Since C language is used to write the programs, the software can be easily integrated with various sensors and drivers of actuators for real application in the future.