Auction-based task allocation for teams of self-reconfigurable robots

Abstract

Self-reconfigurable robots are versatile machines composed of many small, computationally independent modules. If a large number of modules is available, they can divide into multiple smaller robots instead of remaining as one large group. In such a case, they must decide based on their mission whether and when such a division is appropriate, and if so, how to divide the mission objectives amongst the smaller robots. Likewise, after division, a later task may be more effectively handled by merging two groups of modules to obtain a single robot with greater capability. For traditional teams of mobile robots, auction-based methods have been used for task allocation with good success. Here we develop an auction approach with new types of bids that incorporate splitting and merging of these robots. This allows the overall system to exploit its ability to vary the number of robots and their capabilities alongside allocation of the overall set of tasks. We apply this protocol to an exploration scenario, show its correctness and describe the results of simulation in several different environments.