A comparison of algorithms for teams of robots

Abstract

The effective use of teams of robots to accomplish large tasks is an important goal of robotic research.. An important part of achieving this is to understand the trade-offs between various approaches to team organizations and communications. The goal of our research was to understand and compare, in a real-world environment, the strengths and weaknesses of a variety of common approaches to multiple-robot algorithms. We have built a robot simulation environment. The behavior of our simulated environment is validated by comparison with the behavior of our real robots. We have used the standard search-and-tag problem for our initial investigation in which a number of targets that must be located in an unstructured environment. A group of robots explore the environment and report the location of the targets. The efficiency of the approaches is compared in terms of time required to locate the first target and the average time to locate the targets that were found. Our initial experiments revealed interference among the robots, the targets and the obstacles. The effect of this interference was larger than expected.