Abstract: An intelligent course planning method and controller for unmanned surface vehicle are provided, an overall optimization from starting point to end point of course deflection angle is achieved based on an optimization principle of approximate dynamic programming; an optimization of minimization of the course deflection angle and a control signal is achieved based on a quadratic form cost function and a performance index by designing a virtual radial basis function neural network and a least square method, a target course expression is obtained, and a stability and a degree of convergence are ensured through a positive-definite constraint of the Hessian matrix of the performance index. Compared with the related art, an overshoot of course deflection and the control signal is reduced, an optimization of flight and steering energy consumption is achieved, and completely data-driven for intelligent course planning for unmanned surface vehicle and a high-accuracy feedback adjustment are achieved.

Abstract: An intelligent course planning method and controller for unmanned surface vehicle are provided, an overall optimization from starting point to end point of course deflection angle is achieved based on an optimization principle of approximate dynamic programming; an optimization of minimization of the course deflection angle and a control signal is achieved based on a quadratic form cost function and a performance index by designing a virtual radial basis function neural network and a least square method, a target course expression is obtained, and a stability and a degree of convergence are ensured through a positive-definite constraint of the Hessian matrix of the performance index. Compared with the related art, an overshoot of course deflection and the control signal is reduced, an optimization of flight and steering energy consumption is achieved, and completely data-driven for intelligent course planning for unmanned surface vehicle and a high- accuracy feedback adjustment are achieved.