Motion Planning for Multi-robot Coordination on Representation Space

Abstract

The multi-robot coordination task is investigated in its Representation Space (RS). First, the RS model of the robot system as well as its prescribed task is formulated by clarifying the internal and external constraints affecting task realization. All constraints are depicted as unreachable areas in RS model. Then, whether a task is feasible or not is transformed to check 1) if the final representation denoting task realization of the robot system is reachable; and 2) if there is a connecting trajectory illustrating the process that the start representation varies to the final representation, amidst those unreachable areas denoting system constraints. Performance of different planning algorithms are also evaluated upon the representation space, from which the optimal planning algorithm could thus be recognized. The task of multi-robot formation navigation in a warehouse environment is exemplified to illustrate the performance of the proposed scheme: the formation of the robot system is maintained in motion with collisions avoided. Moreover, by incorporating the task oriented motion planning framework, it is capable of transforming an infeasible task into a feasible one with least adjustments in the system's formation to adapt to inner and/or external constraints during task realization.