Path Planning of AUV Swarms Using a Bio-inspired Multi-Agent System

Abstract

This paper uses a multi-agent system (MAS) as the bioinformatics-inspired technique for guiding a team of autonomous underwater vehicles (AUVs) toward the desired destination. An individual AUV is designated as an agent connected by a communication network and assumes full communication. Here, each AUV estimates the position of its neighbor AUVs while moving toward the destination. The proposed multi-AUV system consists of a leader AUV and five follower AUVs. A distributed path consensus (DPC) is proposed to ensure the neighboring agent AUVs maintain a predefined distance between each other while moving toward the predefined destination. Due to the proposed distance constraint between neighboring agents, AUVs stay at a safe distance from each other while maintaining underwater communication using interactive switching topology. The performance of the optimized path is obtained using MATLAB simulation. The proposed algorithm is applied in both, formation using the desired shape, and trajectory tracking, and found to be globally asymptotically stable. The results of the simulation confirm that each agent switches from one state to another and progress over time until the desired coordinated shape is achieved without inter-vehicular collision. The proposed method solves coordination problems among multiple AUVs and increases the coverage of underwater missions like oceanographic surveys.