Discrete Partitioning and Coverage Control for Gossiping Robots

We propose distributed algorithms to automatically deploy a group of mobile robots to partition and provide coverage of a non-convex environment. To handle arbitrary nonconvex environments, we represent them as connected graphs. Our partitioning and coverage algorithm requires only shortrange, unreliable pairwise "gossip" communication among the agents. The algorithm has two components: (1) a motion protocol to ensure that each robot communicates with its neighbors at least sporadically, and (2) a pairwise partitioning rule to update territory ownership whenever two robots communicate. By studying an appropriate dynamical system on the space of partitions, we show that territory ownership converges to a centroidal Voronoi partition in finite time. Additionally, we characterize the scalability properties for the algorithm and describe how it can be implemented on robots with limited computational resources. Finally, we report on large-scale simulations in complex environments ...

• Type: Journal
• Year: 2010
• Where: CORR
• Authors: Joseph W. Durham, Ruggero Carli, Paolo Frasca, Francesco Bullo
• Keywords: Algorithm, CORR 2010, Education, Robot, Territory Ownership