This paper combines probabilistic roadmaps with potential fields in order to enable a robotic swarm to effectively move to a desired destination while avoiding collisions with obstacles and each other. Potential fields provide the robots with local, reactive, behaviors that seek to keep the swarm moving in cohesion and away from the obstacles. The probabilistic roadmap provides global path planning which guides the swarm through a series of intermediate goals in order to effectively reach the desired destination. Random walks in combination with adjustments to the potential fields and intermediate goals are used to help stuck robots escape local minima. Experimental results provide promising validation on the efficiency and scalability of the proposed approach. Source code is made publicly available.
展开▼