Abstract: ObjectiveAiming at the path following and obstacle avoidance problems of underactuated ships with model uncertainty and external environmental disturbance, an adaptive finite time sliding mode control law based on radial basis function neural network minimum learning parameter (RBFNNMLP) algorithm and an improved artificial potential field are proposed by combining command filter technology, minimum learning parameter (MLP), and simulated annealing algorithm.MethodsFirstly, based on the guidance of the virtual ship, a virtual control law is designed according to the following error and azimuth angle. The command filter technique is introduced to estimate the derivative of the virtual control law, reducing the computational complexity. Combining algorithm of RBFNNMLP, a finite time sliding surface is designed for control input, and single parameter online learning is used instead of ownership value online learning to avoid the problem of dimensionality explosion. Control law and adaptive law are also designed. Lyapunov Stability analysis proves that the system is stable in finite time. Improve the repulsion function of artificial potential field for static and dynamic obstacles separately, and introduce simulated annealing algorithm to overcome the local minimum problem and the problem of not considering the position and relative velocity relationship between ships and obstacles.ResultsThe simulation comparison results show that under wave interference, the designed controller has higher convergence accuracy, shorter convergence time, and can effectively avoid obstacles when the ship falls into local minima, verifying the effectiveness and robustness of the proposed control algorithm.ConclusionsThe proposed algorithm can provide reference for underactuated ship path following and obstacle avoidance problems.