Improving particle swarm optimization algorithm for distributed sensing and search

Distributed coordination is critical for a multi-robot system in collective cleanup task under a dynamic environment. In traditional methods, robots easily drop into premature convergence. In this paper, we propose a swarm-intelligence based algorithm to reduce the expectation time for searching targets and removing. We modify the traditional PSO algorithm with a random factor to tackle premature convergence problem, and it can achieve a significant improvement in multi-robot system. It performs well even in a obstacle environment. The proposed method has been implemented on self-developed simulator for searching task. The simulation results demonstrate the feasibility, robustness, and scalability of our proposed method than previous methods.