Swarm Robots Search Under Conditions of Heterogeneous Sensory Signals Fusion.

Abstract

Particle swarm optimization (PSO) is extended to model swarm robots for search on the gas outburst spot in coal mines. The fusion of intermittent cry-for-help, periodic RF wave and continuous gas is employed to guide robots to approach victim. We introduce detecting-success binary logic and sensing event into describing the sense process. Based on statistical properties, detecting range, localization type and estimate precision of signals, the decision sensor is determined with information entropy. When gas is chosen, we control swarm with extended PSO model; while either RF wave or cry is selected, we estimate target location with received signal strength indicator method and modify the model on-line. Finally, simulation is conducted in closed signal propagation environment, set with models of indoor acoustic energy decay, log-normal radio waves and Gaussian plume. The results indicate the validity of the fusion method and control strategy presented.