Haihang Sun and Robert McCartney, University of Connedticut, USA
Task allocation and load balancing are critical in faulttolerant robotic team design. This paper deals with the problems of allocating and reallocating tasks to multiple robots in a fault-tolerant manner. Using a tree structure to store global information about active robots and unfinished sub-tasks, each robot can reallocate itself to tasks independently when robot failure occurs in the team. Organized in full distributed manner, robots in the team respond robustly, flexibly and without negotiation with other team members to robot failure and other environment changes. Simulation results are given to show the feasibility of this approach.