一种基于小脑模型关节控制器评论-策略家的机器人跟踪控制算法

本文提出了一种基于小脑模型关节控制器(CMAC)的评论–策略家算法,设计不依赖模型的跟踪控制器,来解决机器人的跟踪问题.该跟踪控制器包含位置控制器和角度控制器,其输出分别为线速度和角速度.位置控制器由评价单元和策略单元组成,每个单元都采用CMAC算法,按改进δ学习规则在线调整权值.策略单元产生控制量;评判单元在线调整策略单元学习速率.以双轮驱动自主移动机器人为例,与固定学习速率CMAC做比较,仿真数据表明,基于CMAC的评论–策略家算法的跟踪控制器具有跟踪速度快,自适应能力强,配置参数范围宽,不依赖数学模型等特点.

Robot tracking control algorithm based on cerebella model articulation controller adaptive critics and strategists method

The adaptive critic and associative search algorithm based on cerebella model articulation controller(CMAC) is presented and the corresponding controller without referring to the system model is designed to solve the tracking problem of the robot. The controller consists of the position tracking controller and the angular tracking controller; the outputs of which are linear velocity and angular velocity. Either the position tracking controller or the angular tracking controller includes the associative search unit and the adaptive critic unit. Both units are based on the CMAC algorithm and employ the modified ±-learning rule for adjusting online the weight values. The associative search unit is the main part of the controller; and the adaptive critic unit adjusts the learning speed of associative search CMAC. A two-wheeledmotor mobile robot is taken as the example in simulations. Compared with the fixed learning speed CMAC, the proposed controller provides faster tracking response, better adaptive ability and wider parameter range, and is independent of the model.