A method for robot control is proposed in this paper to suppress human-caused disturbance in human-robot collaborative manipulation. In the method, the robot control algorithms are chosen according to the impacts of human motion on the manipulated objects: the modified model predictive controller is used when the impact is large, and the impedance controller is used when the impact is small. In the modified model predictive control, the human motion in a specific direction is considered as disturbance, the disturbance is observed, predicted, and its impact on the manipulated objects’ stability is estimated. The robot control parameters are then optimized based on the estimation. A series of simulation and physical experiments are conducted. The results show that the modified model predictive control shows better stability than the impedance control and model predictive control. Specifically, the maximum displacement of the manipulated objects decreases by 70% compared with the impedance control and 44% compared with the model predictive control.

本文提出了一种机器人控制方法，以抑制人机协同操作中人为干扰。该方法根据人体运动对被操纵物体的影响来选择机器人控制算法：当影响大时采用修正模型预测控制器，当影响较小时采用阻抗控制器。在修正模型预测控制中，将特定方向的人体运动视为扰动，对扰动进行观察、预测，并估计其对被操纵物体稳定性的影响。然后基于估计优化机器人控制参数。进行了一系列的模拟和物理实验。结果表明，改进的模型预测控制比阻抗控制和模型预测控制表现出更好的稳定性。具体而言，被操纵物体的最大位移与阻抗控制相比减少了 70%，与模型预测控制相比减少了 44%。