A reinforcement learning-based global coverage method for unmanned clusters, program, device and storage medium

Abstract

The present application belongs to the technical field of underwater acoustic detection, and particularly relates to an unmanned cluster global coverage method based on reinforcement learning, a program, an equipment and a storage medium. The present application constructs the global coverage task of the unmanned cluster in the target sea area as a distributed partially observable Markov decision process, constructs a detection and sensing map with a revisit mechanism, quantitatively represents the coverage distribution state of the global sea chart, and effectively guides the intelligent emergence behavior of the water surface unmanned cluster, i.e. repelling the densely covered areas and being attracted to the unknown or low attention areas. The present application designs the reinforcement learning framework elements suitable for the global coverage task according to the special requirements of the global coverage task of the water surface unmanned cluster, embeds the complex acoustic physical constraints and space-time state into the mathematical model of the reinforcement learning, greatly reduces the exploration difficulty of the intelligent agent in the complex environment, and accelerates the convergence process of the MATD3 algorithm.

Images