A team of "PrairieDog" robots

We are interested in designing distributed algorithms that allow multiple robots to pool their resources for problem solving in realistic communication environments.

When multiple robots are faced with a common problem that is computationally intensive (e.g. coordinated navigation and/or task allocation), it is advantageous if all affected robots can contribute resources to help find a solution. Dividing the combined effort among many robots allows better solutions to be found more quickly.

We look at how this type of cooperation can be achieved in real environments. Therefore, a major design consideration is the ability to function given imperfect or poor communication between the individual computational nodes. We are also interested in how to exploit good communication when it exists. We call this type of algorithm “Any-Com” because it automatically adjusts to the available communication quality.

A multi-robot path planning problem

One strategy we have explored is to have the robotic team find an initial (possibly suboptimal) solution and then refine it toward optimality as communication permits. In this framework, successful communications allow better solutions to be found more quickly but unsuccessful communications do not prohibit a solution from being found.

Applications of such algorithms range from coordinating robots in warehouses, precision agriculture to cars on a parking lot or in city traffic.

Publications

M. Otte and N. Correll. Any-Com Multi-Robot Path Planning: Maximizing Collaboration for Variable Bandwidth. In The 10th Int. Symp. on Distributed Autonomous Robotic Systems (DARS), Lausanne, Switzerland, November 2010. To appear.

M. Otte and N. Correll. Any-Com Multi-Robot Path-Planning: Multi-Robot  Coordination under Communication Constraints and Dynamic Team Sizes. In International Symposium on Experimental Robotics (ISER), New Delhi, India, December 2010. To appear.

M. Otte and N. Correll. The Any-Com Approach to Multi-Robot Coordination. In IEEE International Conference on Robotics and Automation (ICRA), Workshop on Network Science and Systems Issues in Multi-Robot Autonomy (NETSS), Anchorage, USA, 2010.

 

Leave a Reply

Set your Twitter account name in your settings to use the TwitterBar Section.