Multiple Source Seeking using Glowworm Swarm Optimization and Distributed Gradient Estimation

Abstract

In this paper we propose a novel mission control strategy for a group of agents to locate unknown multiple extrema of a time-invariant scalar field. The goal is to provide a complete solution for realistic scenarios such as malfunction of agents or time-varying topology. Here we present MESA - Multiple Extrema Seeking Algorithm, where a group of autonomous agents are able, from any arbitrary initial location (in a bounded area), to locate multiple, unknown extrema. The proposed algorithm includes a combination of bio-inspired attraction with estimated gradient and formation control, where the task switching methodology is performed under the specification of Linear Temporal Logic formula. At first, agents find neighbors and construct groups for completing the search procedure. Once an extremum is 'located', a virtual repulsive force is produced by the pioneer group which as a result cause others to search for other unexplored extrema. We propose a new measure-source density - evaluating the algorithm efficiency which is tested under different simulations for a unicycle agent's model.