FF-RRT: A Sampling Based Path Planner for Flexible Multi-Robot Formations

Abstract

In this paper we present a centralised path planning approach, called flexible formation rapidly exploring randomized trees (FF-RRT) algorithm for navigation of formations of multiple homogeneous robots in obstacle-ridden environments. The path for the multiple robots, in a flexible formation, is planned in a configuration space, defined by its translation, rotation and scaling. The path of individual robots are then found using affine transformations of configurations in the planned path. The flexible formation was able to overcome obstacles that were impossible for rigid formations successfully by changing its size and orientation along the path. For large obstacles split and merge type behaviour was observed for the flexible formation, which would not be possible for rigid formations.