Task Scheduling in Multi-robot Systems: An Improved Consensus-Based Bundle Algorithm

This paper proposes a novel consensus-based autonomous grouping algorithm to address a task scheduling problem for a fleet of autonomous robots. As an extension of the consensus-based bundle algorithm, the proposed algorithm is capable of dealing with not only the single-robot tasks but also the tasks requiring the cooperation of a specific number of robots, i.e., the so-called multi-robot tasks. The proposed algorithm is executable based on local communication among neighbor robots as long as a strongly connected communication network is given. Further, some critical problems, including time synchronization and task exit, are analyzed in detail. The effectiveness and superiority of our algorithm is verified through in-depth numerical simulations.