Discrete learning control for robots: strategy, convergence and robustness

摘要

Iterative learning control serves to enhance the performance of robots carrying out repetitive cycles of operation. A new discrete learning method is developed which (a) is computationally not too demanding since the full model of the robot is not involved and (b) can reduce design tuning to simply choosing the sampling interval in order to provide satisfactory performance. The convergence issue and robustness problem are particularly addressed. The effectiveness of the learning control method is demonstrated by experimental results obtained using a commercial robot.