A swarm intelligence based algorithm for distribute search and collective cleanup

A collective cleanup task requires a multi-robot system to search for randomly distributed targets and remove them under a dynamic environment. In traditional methods, robots wandered in subareas (which caused too much repeat search) and interchanged all detected information with their neighbors, so global searching time and communication traffic increased. In this paper, we propose a swarm intelligence based algorithm that minimizes the expected time for searching targets by dividing the environment into two levels subareas then using a dynamic computing subareas' probability algorithm for search strategy, and it can also reduce communication traffic by robots' selective information interactions with their neighbors. A modified Particle Swarm Optimization (PSO) method is used to balance searching and selecting, which helps to allocate reasonable robots to different targets. The simulation results demonstrate the higher efficiency of the proposed method when compared to another method.