含未知信息的轮式移动机器人编队确定学习控制

Formation control of wheeled mobile robots with unknown information via deterministic learning

中文关键词  未知信息  移动机器人编队  非完整约束  系统动态  学习控制

英文关键词  unknown information  mobile robot formation  nonholonomic constraint  system dynamics  learning control

中文摘要

      本文研究含未知信息的轮式移动机器人(wheeled mobile robots, WMR)的编队控制问题. 首先, 基于领航– 跟随法和虚拟结构法, 将WMR编队控制问题转化为跟随机器人对参考虚拟机器人的跟踪控制问题. 然后, 利用径向 基函数神经网络(radial basis function neural networks, RBF NN)对WMR的未知系统动态进行学习, 以及根据李雅普 诺夫稳定性理论设计了稳定的自适应RBF NN控制器和RBF NN权值估计的学习率. 依据确定学习理论, 闭环系统 内部信号在对回归轨迹实现跟踪控制的过程中满足部分持续激励(persistent excitation, PE)条件. 随着PE条件的满 足, RBF NN权值估计收敛到其理想权值, 实现了对未知闭环系统动态的准确学习. 最后, 利用学习结果设计了RBF NN学习控制器, 保证了控制系统的稳定与收敛, 实现了闭环稳定性和改进了控制性能, 并通过仿真验证了所提控制 方法的正确性和有效性.

英文摘要

      This paper investigates the formation control of wheeled mobile robots (WMR) with unknown information under nonholonomic constraints. Firstly, based on the leader-follower method and the virtual structure method, the formation control is transformed into the problem that the followers track their virtual leader. Secondly, a radial basis function neural network (RBF NN) is used to learning the unknown information (closed-loop system dynamics) of WMR, and a stable adaptive RBF NN controller and the stable adaptive tuning law of RBF NN parameters are derived in the sense of the Lyapunov stability theory. According to deterministic learning, a partial persistent excitation (PE) condition of some internal signals in the closed-loop system is satisfied in the control process of tracking a recurrent reference trajectory, and an accurate approximation of the unknown closed-loop system dynamics is achieved by the RBF NN parameters convergence to their optimal weights. Finally, a RBF NN learning controller which effectively utilizes the learned knowledge without re-adapting the RBF NN parameters is proposed to achieve the closed-loop stability and improve the control performance, and simulation studies are included to demonstrate the correctness and effectiveness of the proposed approach.