Adaptive sampling and collective behaviour in a small team of AUVs for a source localization problem

In this paper we present an adaptive sampling approach for a distributed multi-AUV source localization algorithm. The algorithm is composed of different bio-inspired behaviours for invoking collective behaviour in a multi-agent system. The adaptive sampling approach shows significant performance improvement in terms of mean arrival times over the previously reported static sampling approach for both the presented arrival scenarios. The adaptive sampling approach also boosts the maximum levels of team cohesion, suggesting its ability to maximize the benefit from the social information. Apart from better mean arrival times, the adaptive sampling approach also results in significantly reduced variance of the arrival-time distribution.