Collective-adaptive Lévy flight for underwater multi-robot exploration

This paper presents the use of Lévy flight, a bio-inspired algorithm, to efficiently and effectively locate targets in underwater search scenarios. We demonstrate how a novel adaptation strategy, building on the Firefly optimization algorithm, substantially improves Lévy flight performance. The adaptation strategy represents a swarm intelligence approach, the distribution patterns governing robot motion are optimized in accordance with the distribution of targets in the environment, as detected by and communicated between the robots themselves. Simulation experiments contrasting the performance of the present Lévy flight and two other search strategies in both sparse and clustered distributions of targets are conducted. We identify Lévy flight as exhibiting the best performance, and this is improved with our adaptation strategy, particularly when targets are clustered. Finally, Lévy flight's superior performance over the alternative strategies examined here is empirically confirmed through deployment on real-world underwater swarm robotic platforms.