Multi-robot Planning for Non-overlapping Operator Attention Allocation

摘要

As society increases its reliance on remotely-operated and partially autonomous robots, we face the problem of how to effectively interact with large numbers of robots given a limited number of operators. In this paper, we present a methodology to allocate operator attention across multiple robots before plan execution while minimizing the duration of the plan. Our methodology has the following steps: 1) accept trajectories for robots and analyze where they will require supervision, 2) create a coordination space representing operator attention conflicts, 3) apply algorithms to attempt to reduce the time-span of robot activity by devising control policies, or assigning alternate trajectories, and 4) return a policy for each robot representing a feasible solution where operator attention is not overextended. Finally, we present software simulations and a physical implementation to demonstrate the usefulness of the methodology.