Model-Free Data-Driven Prescribed-Time-Performance Control for Underactuated ASV

In this paper, a model-free data-driven prescribedtime-performance (DDPTP) algorithm is developed for the underactuated autonomous surface vehicle (ASV). By defining a leader to provide the dynamic reference signal, the proposed DDPTP algorithm can achieve high-precision target tracking tasks while ensuring collision avoidance and connectivity between the leader and follower. A prescribed-time-performance function (PTPF) is designed for the ASV, featuring an adaptive shape and customizable convergence time. By using both real-time and historical data, a data-driven neural predictor (DDNP) is developed to estimate system uncertainties including unmodeled dynamics, unknown environmental disturbances and unknown input gains. An adaptive control law is employed based on the knowledge obtained from the DDNP, without relying on prior information about the motion model parameters. A switching dynamic event-triggered mechanism (SDETM) is introduced, which can switch between fixed-threshold triggering mechanism and dynamic event-triggered mechanism (DETM). Through this SDETM, not only are communication resources conserved, but frequent updates of the control signal are also mitigated. The effectiveness of the proposed algorithm is validated through verification simulations.