Research on Task Assignment Based on Hierarchical Structure for Autonomous Underwater Vehicle

Abstract

There are two structures in formation control system for Autonomous Underwater Vehicle (AUV), such as centralized and distributed structures. Centralized cluster formation brings too much overhead to the master node, and the distributed cluster formation poses a huge challenge to collaborative communication. To address these problems, this paper proposes a hierarchical structure for AUV cluster formation task allocation strategy. The strategy deals with the local cluster information hierarchically, uses priority sampling to sample the hierarchical cluster information and stores it in the experience pool, and then uses the Actor-Critic dual network for evaluation to obtain the optimal strategy. The experimental simulation results show that the method proposed in this paper is consistent between the master and the slave, and the system improves the convergence speed with a smaller number of iterations.