Distributed Data-Driven Formation Control for Unmanned Vehicle Networks Under Switching Topologies

Unmanned vehicle teams are envisaged to have broad application prospects and play a key role in future various cooperative engineering projects. This paper studies the formation control protocol design problem via input-output system data for multiple underwater vehicle networks involving highly nonlinear dynamics, strong couplings between the rotational motion and the translational motion, and switching topologies. A distributed data-driven formation control protocol based on reinforcement learning is developed for multiple underwater vehicles to achieve the translational formation and rotational synchronization with the leader, without the requirement of the vehicle dynamics information. The optimality of the proposed formation controller for UUVs is achieved via the reinforcement learning approach. Simulation results for a multi-vehicle network demonstrate the effectiveness of the developed data-driven formation control protocol.