Application of CMAC-based algorithm of critics & strategists on robot tracking control

The CMAC-based algorithm of critics & strategists is proposed in this article for application on robot tracking control. This tracking controller comprises position tracking controller and angle tracking controller, whose outputs are line rotation speed control value and angle control value respectively. Both of them include in two CMAC controllers dealing with online weight value adjustment through the δ learning rule. The strategy CMAC is used to generate control quantity, while the evaluation CMAC is used to perform online adjustment for the strategy CMAS on the basis of controlled errors. Taking double-wheel driven autonomous mobile robot as an example, comparing with the fixed learning rate CMAC, the simulation data indicates that the tracking controller based on CMAC reinforcement learning algorithm features high tracking speed, strong self-adapting ability, wide range of configuration parameters, and independence of mathematical model etc.