Control for Physical Human-robot Interaction Using Dynamics with Successive Iteration of Vector Field

就服務機器人與人的身體交流研究而言，保持他們各自運動的同步是相當重要的。為使該類機器人保持與人運動的同步，本論文提出了動力學矢量場逐次疊代設計的人和機器人運動控制模型。首先，在三維空間裏設計了具有任意吸引子的動力學逐次疊代矢量場，該矢量場能實現機器人關節扭矩輸入信號和關節位移輸出信號的同步，且通過調節遺忘參數和同步闔值可改變同步強度。其次，分析了兩個不同矢量場相互作用時的輸入輸出信號的同步情況。最後，利用所提出的控制模型進行人和7自由度機器人手臂的握手實驗。同時，為便於實時控制，通過參數識別和BP-NN神經網絡對該機器人臂進行重力補償，將被償後的扭矩作為控制模型的輸入信號，並將控制模型的輸出作為機器人關節的期望位移。實驗結果表明，該控制模型對於協調同步人和機器人的相互運動是有效的。

關鍵字

無資料

並列摘要

Physical human-robot interaction(PHRI) is an important subject for the service robots. To make a natural physical interaction, synchronization and entrainment between a robot motion and a human motion play an important role. From this point of view, a control model adopting dynamics with successive iteration of vector field is proposed for PHRI. Firstly, the vector field of the dynamics with arbitrary attractor is designed by adopting successive iteration method in 3-D space, which aims to realize the synchronization between input and output signals, and the strength of the synchronization can be varied by adjusting the oblivion parameter and threshold value in the dynamics. Secondly, the mutual interaction between two different dynamics is analyzed for synchronization between input and output signals. Lastly, based on 7 DOF robot arm, experiments are performed for human-robot handshaking with this control model, and the input and output of the dynamics is compensation torque and desired displacement of each joint respectively. In order to realize the real-time control, gravity compensation for the robot arm is implemented with back propagation neural network (BP-NN) learning and parameter identification methods. The experiment results indicate the model’s validity to synchronization between robot and human motions.

並列關鍵字

successive iteration design ； dynamics ； gravity compensation ； parameter identification.

國際替代計量

Control for Physical Human-robot Interaction Using Dynamics with Successive Iteration of Vector Field

全文下載

主題瀏覽