A Minimalistic and Decentralised Approach to Formation Control for Crowded UUV Swarms Inspired by Fish Schooling

Abstract

Abstract In this paper, we propose a novel approach to formation control for a swarm of unmanned underwater vehicles (UUVs) inspired by the collective behaviour of fish schools. The approach addresses the challenges of centralised control, inflexibility, and high computational complexity in existing formation control methods. The proposed strategy for an individual agent consists of three parts: (1) coordinated strategy of collective motion without a predefined target, in which each individual agent adjusts its motion by rotating a fixed angle to the left or right based on the observed motion direction of its neighbours; (2) target guidance policy that the motion direction of each agent is guided by the target area; and (3) dispersion control strategy of avoiding overcrowding near the target area, in which each individual determines its own target position based on the relative positions of its neighbours and the target area. The simulation experiments demonstrate that using the proposed method, a group of individuals can perform an area coverage search in a flexible collective motion manner by simply making a left or right movement based on their own local perception. Furthermore, we validated the effectiveness of the proposed method using a UUV swarm consisting of 9 UUVs in a UUV simulator. The proposed method is flexible, decentralised, and has low computational complexity. Therefore, it can be applied to the formation control of large swarms of UUVs with limited perception and decision-making abilities.