Learning-based Near-Optimal Formation Control of Wheeled Mobile Robots with Input Saturations

This paper investigates the formation control problem of wheeled mobile robots (WMRs) in the case of actuator input saturations. A learning-based distributed near-optimal formation control algorithm for WMRs with the kinematic and dynamic model is proposed. In the algorithm, the kinematic controllers of WMRs in the formation are obtained by the distributed backstepping technique, which generates reference signals for the dynamic model of the WMRs. Then, the learning-based dynamic controllers of the WMRs in the formation are designed by the actor-critic-based reinforcement learning algorithm. To deal with the problem of the control input saturations, the performance index with a non-quadratic function is designed. Subsequently, neural networks are employed to approximate the near-optimal costate functions and control policies in the learning-based dynamic controllers of WMRs in the formation. Finally, various numerical simulations are performed to validate the effectiveness and feasibility of the learning-based formation control algorithm.