Abstract
<p>The development of a low cost and intelligent swarm of autonomous underwater vehicles (AUVs) is the long term goal of this research. Such a swarm of AUVs could be used, for example, for locating submarine sources of interest, such as dumped radioactive waste or ammunition. This is a difficult problem for direct search algorithms, because in large areas of the search space, gradient information is not available. The overall search strategy used in the work is based on particle swarm optimisation (PSO). The number of AUVs here is small due to costs and availability. The performance of PSO depends on the right choice of control parameters. Therefore this paper presents an empirical study of the effects of different search parameter settings on the performance of PSO, used with a swarm of three AUVs in a dynamic environment. A simulation of submarine groundwater discharge, based on Cellular Automata, is used as a dynamic test environment. It was shown in this research that PSO in this configuration is robust against control parameter settings. <br></p>