Abstract
A learning control method, which generates feedforward inputs described by B -spline curves, is proposed. The algorithm is composed of two parts: Local Learning (LL) and Global Learning (GL). LL iteratively improves a feedforward input to approach the system output to a specified desired trajectory. GL identifies the controlled system dynamics by trial movements. The method has following features: (1) Describing inputs as B-spline curves has two merits: Input trajectories obtained by LL need relatively less meories to be stored and they are useful for the feedback control because of their differentiability. (2) LL is expressed by a simple iterative rule. (3) The gain matrix of the iterative rule is determined by GL. (4) GL is useful in the meaning that the identification process does not need any knowledge about the number of parameters and the order of dynamics. (5) The inverse dynamics obtained by GL generates the input trajectory which is a good initial trial for LL. The proposed method is effectively applied to the tracking control for a one-link flexible arm.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
