Research on Anti-pursuit Evasion Strategy of Unmanned Surface Vehicle Based on T-D3QN

From the perspective of the evader in the pursuit evasion problem of the unmanned surface vehicle (USV), this paper proposes an anti-pursuit evasion strategy for the USV based on Twin-Dueling Double Deep Q Network (T-D3QN), which uses two independent Q networks to estimate the Q value. According to characteristics of the pursuit-evasion problem of the USV, the research designs state space, action space and reward function and constructs the simulation environment to train the model. Simulation experiments show that the performance of T-D3QN is better than DQN because it can improve the escape success rate of the USV, and verify the effectiveness of the evasion strategy for the USV.