Optimized Backstepping Strategy for Tracking Control of unmanned surface vehicles

This study addresses the control problem of achieving accurate path following for underactuated unmanned surface vehicles (USVs) by integrating backstepping and optimal control techniques. Neural network (NN) approximators are employed to facilitate the solution of the Hamilton-JacobiBellman (HJB) equation for achieving optimal tracking performance. Since an exact solution of such equation is intractable, an actor-critic reinforcement learning framework is adopted to estimate the optimal control policy and the corresponding cost function. The Lyapunov stability theorem is used to demonstrate that the tracking errors remain semiglobally uniformly ultimately bounded. Simulations are presented to verify the efficacy of the proposed approach.