操作行为估计的人机交互控制实验

Behavior Estimation Based Control for Physical Human-Robot Interaction

为提高物理人机交互系统的柔顺效果,采用引入径向基神经网络进行物理人机交互(pHRI)过程中操作行为估计,以有限时间稳定的非奇异终端滑模为基础,改进对数项以提高pHRI过程中跟踪误差在面上运动的收敛能力,并部署操作人员和触觉力反馈装置的交互控制实验,完成控制理论和机器人学综合创新拓展实训环节。实验表明,在pHRI过程中,基于操作行为估计的人机交互控制方法能够实现人机柔顺交互,并且保持0.3 mm的跟踪误差,所设计方法的具有有效性。

Aiming to enhance the compliance performance in the physical human-robot interaction(pHRI)system,a radial basis function neural network is synthesized into the closed-loop system to estimate the operation behavior during pHRI.Based on the non-singular terminal sliding mode with finite-time stability satisfied,the improved logarithmic term is employed to improve the convergence of tracking error during its motion.The experiment of pHRI with operator and tactile force feedback devices are deployed to complete the comprehensive innovation and expansion training of control theory and robotics.Experimental results show that in the pHRI the proposed behavior-estimation based control method can make the human-robot interaction flexible and maintain a tracking error of 0.3 mm,which proves the effectiveness of the designed method.