Cooperative learning control of uncertain nonholonomic wheeled mobile robots with state constraints

This article investigates the cooperative tracking control of multiple homogeneous uncertain nonholonomic wheeled mobile robots with state constraints. Transforming each mobile robot system into a chained form, a cooperative learning control scheme based on the adaptive neural network is proposed. Firstly, a virtual control law is designed for the kinematic model of the constrained chain system combined with the barrier Lyapunov function (BLF). Then, radial basis function neural networks (RBF NNs) are exploited to deal with the unknown nonlinear dynamics in the mobile robot system, a robust term is introduced to compensate for the NN approximation errors and the external disturbance, and the Moore–Penrose inverse is adopted to avoid the violation of state constraints. Communication network is used to realize the online sharing of NN weights of each mobile robot individuals, such that locally accurate identification of the unknown nonlinear dynamics with common optimal weights can be obtained. As a result, the learned knowledge can be reused in the cooperative learning control tasks and the trained network model has better generalization capabilities than the normal decentralized learning control. Finally, numerical simulation verifies the effectiveness of the control scheme.