Multi-source signals guiding swarm robots search

Abstract

Victim search with swarm robots on gas outburst spot in coal mines is investigated. Upon modeling swarm robotic system with extended particle swarm optimization, heterogeneous signals are fused by each robot independently as fitness evaluate. To fuse signals of intermittent sound of call for help, periodic radio frequency waves and continuous gas, concepts of detection-success binary logic and perceptual event are introduced into describing detection process. Based on statistical properties, detectable range, localization type and accuracy, priority is determined with information entropy-based criteria. Then the best-found position is decided in a distributed fashion through election operation. Simulation conducted in closed signal propagation environment indicates the approximate relation between fusion value and distance, i.e., the nearer the robot is far away from target, the higher the fusion of signals. © 2009 Springer Berlin Heidelberg.