On the techniques of multi-agent path planning and the collaboration

Beginning with the current situation of study on the multi-robot system, the paper introduces the feature and structure of multi-robot system and presents the planning methods and characters about the static and dynamic path. Following the above analysis, this paper puts forward and realizes the collaborating of task-level and harmonizing of locomotion-level in the multi-robot system. The path planning structure of “global layout and local amendment” is taken and a strategy of clash relieving based on forecast is designed as the local programming method. What´s more, the Visual C++6.0 is used during the simulated experiments about the path planning and obstacle avoidance to both single-robot and multi-robot motionless obstacle. The experiment results make clear that the method of task assignment could satisfy the real-timing and optimality of motionless task distribution and the autonomy and robustness of dynamic task assignment, and additionally, the measure of clash relieving has the preferable feature of coordinating. Furthermore, the collision strategy based on intensive learning could effectively realize the robot´s initiative programming and enhance its adaptability to environment.