Discrete-time prescribed performance compliant control based on online-learning for human-robot collaboration system
To enable robot compliant to human behavior in human-robot collaboration system accurately,a discrete-time prescribed performance compliant control based on online-learning is proposed.This method employs online sequential extreme learning machine in the outer loop to estimate the human behavior.The estimation results are combined with a reference impedance model to reconstruct the reference trajectory.A discrete-time prescribed performance controller is established in the inner loop to track the reconstructed reference trajectory,and time delay estimation is employed to obtain the unknown dynamics of the robot.The transient and steady performances of the closed-loop system are analyzed.The effectiveness of the proposed controller is verified by comparative simulations.The proposed discrete-time control method can better satisfy the working principle of digital computers,and it can make the tracking error of the end-effector meet the prescribed performance with less computation and memory burden.In addition,the proposed method does not require the accurate mode of the robot,reduces the force burden of human operating the robot,and guarantees the compliance to the human behavior.
万方数据
维普
