In this paper, we explore the implementation of standard particle swarm optimization (SPSO) on a swarm of physical mobile robots conducting a source seeking task. The signal source is electromagnetic, whose strength is non-differentiable at many points making most gradient based source seeking strategies ineffective in this scenario. We analyze the physical limitations of the robots and modify SPSO accordingly to make them compatible with each other. We also compare different SPSO topology models to determine the one best suited for our problem. Finally, we incorporate obstacle avoidance strategies into PSO, and compare the performance of original PSO, SPSO 2006 and SPSO 2011 in a complex environment with obstacles. Simulation results demonstrate the efficacy of implementing SPSO to robot source seeking problem. Moreover, it is shown that SPSO 2011 is not only superior as an optimization method, but also provides better performance in robotic implementation compared to SPSO 2006 and original PSO.
DisciplinesAeronautical Vehicles | Electro-Mechanical Systems | Navigation, Guidance, Control and Dynamics | Robotics | Signal Processing
CommentsThis is a manuscript of a proceeding published as Zou, Rui, Vijay Kalivarapu, E. Winer, J. Oliver, and S. Bhattacharya. "Standard particle swarm optimization on source seeking using mobile robots. In this paper, we explore the implementation of standard particle swarm optimization (SPSO) on a swarm of physical mobile robots conducting a source seeking task. The signal source is electromagnetic, whose strength is non-differentiable at many points making most gradient based source seeking strategies ineffective in this scenario. We analyze the physical limitations of the robots and modify SPSO accordingly to make them compatible with each other. We study the implementation of SPSO in an environment where obstacles and collisions exist and compare the results of this implementation with a basic PSO that was implemented in our prior work.