| Summary: | The leader-follower model is found efficient for robot swarm exploration tasks. When a swarm of mobile robots collaboratively explore an unknown environment with limited communication, such as communication delay or even message loss, it is necessary to adaptively re-group robots according to the dynamic communication conditions and task goals. In this paper, we propose to analyze the leader-follower topology in a graph fashion and employ Ratio Cut of the proposed graph for membership re-assignment. We take each robot as a node and consider both the communication delay and the goal, i.e., exploring towards unknown area, as the edge weight. Then we use the Laplacian matrix to bipartition the graph repeatedly to optimize group partition. Simulation results validate the effectiveness of the proposed method.
|
|---|
