Robustness of P-type learning control with a forgetting factor for robotic motions

A class of simple learning control algorithms with a forgetting factor and a long-term memory and without use of the derivative of velocity signals is proposed for motion control of robot manipulators. The robustness of search learning laws with respect to initialization errors, fluctuations of the dynamics, and measurement noises is studied extensively. As a result the uniform boundedness of motion trajectories is proved based on the passivity analysis of robot dynamics. It is also proved that motion trajectories converge to a neighborhood of the desired one and eventually remain in it provided the content of the long-term memory is refreshed adequately after a sufficient number of trials