基于神经网络反步法的移动机器人路径跟踪控制

Path following controller for mobile robots based on neural network backstepping

为实现非完整轮式机器人的路径跟踪控制,设计基于反馈增益的反步法控制器,通过控制器参数设计消除了机器人动态误差模型中的部分非线性项,采用神经网络对模型不确定项进行补偿,并利用自适应鲁棒控制器在线补偿神经网络的估计误差,优化了神经网络的学习性能。仿真结果表明:设计的控制器参数易于调节,可实现轮式移动机器人对任意曲线路径的精确跟踪。

In order to implement the path following control of wheeled mobile robot with non-holonomic constraint,a backstepping method is designed based on feedback gain technique. Through the tuning of the controller＇s parameters,the nonlinear terms in error dynamic robot model can be e-liminated,and the form of designed controller can be much simpler. Neural network is adopted to compensate the model uncertainties. An adaptive robust controller is designed to compensate the estimated error of neural network on-line,and the learning performance of neural network can be op-timized. The simulation results illustrate that the parameters of controller are easy to be adjusted,and can make wheeled mobile robot track the desired arbitrary path precisely.