Maintaining Connectivity in Mobile Robot Networks

While there has been significant progress in recent years in the study of estimation and control of dynamic network graphs, limited attention has been paid to the experimental validation and verification of such algorithms on distributed teams of robots. In this work we conduct an experimental study of a non-trivial distributed connectivity control algorithm on a team of seven nonholonomic robots as well as in simulation. The implementation of the algorithm is completely decentralized and asynchronous, assuming that each robot only has access to its pose and knowledge of the total number of robots. All other necessary information is determined via message passing with neighboring robots. We show that such algorithms, requiring complex inter-agent communication and coordination, are feasible as well as highly successful in enabling a network of robots to adapt to disturbances while preserving connectivity. Comments Postprint version. Published in: O. Khatib et al. (Eds.): Experi. Robotics: The 11th Intern. Sympo., STAR 54, pp. 117–126. The original publication is available at www.springerlink.com. DOI: 10.1007/978-3-642-00196-3_14 Publisher URL: http://springerlink.com/content/x2358726nt30wu2l/ ?p=e21a7b28945b4ba88dc5bb535c1a79c3π=0 This conference paper is available at ScholarlyCommons: http://repository.upenn.edu/grasp_papers/39 Maintaining Connectivity in Mobile Robot Networks Nathan Michael, Michael M. Zavlanos, Vijay Kumar, and George J. Pappas GRASP Laboratory, University of Pennsylvania, Philadelphia, PA 19104, USA {nmichael,zavlanos,kumar,pappasg}@grasp.upenn.edu Summary. While there has been significant progress in recent years in the study of estimation and control of dynamic network graphs, limited attention has been paid to the experimental validation and verification of such algorithms on distributed teams of robots. In this work we conduct an experimental study of a non-trivial distributed connectivity control algorithm on a team of seven nonholonomic robots as well as in simulation. The implementation of the algorithm is completely decentralized and asynchronous, assuming that each robot only has access to its pose and knowledge of the total number of robots. All other necessary information is determined via message passing with neighboring robots. We show that such algorithms, requiring complex inter-agent communication and coordination, are feasible as well as highly successful in enabling a network of robots to adapt to disturbances while preserving connectivity.