Optimal tracking and event-driven control for USV formations under communication delays and topology switching with input saturation

This paper studies the formation tracking problem for underactuated unmanned surface vehicles (USVs) in asynchronous communication environments induced by time delays and topology switching. A control method based on delay-compensating prediction, reinforcement learning with input saturation constraints and adaptive event triggering is proposed. This method employs a distributed predictor to estimate synchronized states, which drives a topology-aware actor–critic neural network to learn the optimal control law online under actuator constraints. Communication is triggered adaptively based on a reinforcement learning. Theoretical analysis proves the uniformly ultimately bounded stability of the closed-loop system. Simulation results show that the proposed method maintains tracking accuracy under time-varying delays and switching topologies, and reduces the communication load.