1st Workshop on Wireless Multihop Communications in Networked Robotics

Research Article

Autonomous Biconnected Networks of Mobile Robots

Download397 downloads
  • @INPROCEEDINGS{10.4108/ICST.WIOPT2008.3245,
        author={Jesse Butterfield and Karthik Dantu and Brian Gerkey and Odest Chadwicke Jenkins and Gaurav S. Sukhatme},
        title={Autonomous Biconnected Networks of Mobile Robots},
        proceedings={1st Workshop on Wireless Multihop Communications in Networked Robotics},
        keywords={Algorithm design and analysis Artificial intelligence Bandwidth Computer science Intelligent robots Markov random fields Mobile robots Motion analysis Robot kinematics Robustness},
  • Jesse Butterfield
    Karthik Dantu
    Brian Gerkey
    Odest Chadwicke Jenkins
    Gaurav S. Sukhatme
    Year: 2008
    Autonomous Biconnected Networks of Mobile Robots
    DOI: 10.4108/ICST.WIOPT2008.3245
Jesse Butterfield1,*, Karthik Dantu2,*, Brian Gerkey3,*, Odest Chadwicke Jenkins1,*, Gaurav S. Sukhatme2,*
  • 1: Department of Computer Science, Brown University, Providence, RI 02912-1910
  • 2: Department of Computer Science, University of Southern California, Los Angeles, CA 90089
  • 3: Artificial Intelligence Center, SRI International, Menlo Park, CA 94025-3493
*Contact email: jbutterf@cs.brown.edu, dantu@usc.edu, gerkey@ai.sri.com, cjenkins@cs.brown.edu, gaurav@usc.edu


Groups of robots can be used in a coordinated fashion to achieve goals that individual robots cannot. One of the key requirements for this is being able to communicate amongst themselves in a timely and robust manner. This capability has been assumed as available in many multi-robot solutions but has not received enough attention in research. We approach the subproblem of moving from a connected network of robots to achieving biconnectivity. Biconnectivity provides both robustness to change in links and better bandwidth for communication by providing multiple paths to the destination. We take two approaches to the same problem - one a deterministic graph-based approach and the other a Markov random field approach. Both have their advantages. Preliminary results indicate much promise in both these directions.