Study of multi-robot system on task allocation

Abstract

The decentralized multi-robot system is designed for the task allocation as a load balancing problem. For individual robot the behavior-based control methodology and subsumption control architecture,which can maintain a tight coupling between sensing and actions,are used. In distributed multi-robot system each robot uses vision system to make local observations and employs the transition functions so as to choose tasks of desired type to execute,so that the global task allocation can emerge from individual's task switching. Macroscopic mathematical model justifies the effectiveness of task allocation,then the experiments are conducted five times and the results demonstrate that the control exerted to the multi-robot system can attain the desired assignment of tasks.