Zero-sum game based tracking control for fully-actuated AUV with unknown disturbances using critic network

In this work, the trajectory-tracking control of the fully-actuated autonomous underwater vehicle (AUV) with unknown disturbances is formulated as a two-player zero-sum game. One player is the optimal controller which is designed to make the fully-actuated AUV track the desired trajectory. The other player is the unknown disturbances that represent the overall damping effect of the fully-actuated AUV. A single critic network is employed to estimate the performance index, which introduces a discount coefficient due to the nonlinearity and the complexity of fully-actuated AUV. The online policy iteration algorithm and the weight update law are proposed which can achieve the near-optimal control. Using the Lyapunov stability theory, the stability of the fully-actuated AUV is analyzed and the system is guaranteed to be uniformly ultimately bounded (UUB). In addition, the proposed control scheme can relax the requirements for initial admissible control conditions. The effectiveness of the proposed method is demonstrated through the real-time simulations on the fully-actuated AUV model.