Fault-Tolerant and Self-Adaptive Market-Based Coordination Using Hoplites Framework for Multi-Robot Patrolling Tasks

Abstract

An autonomous robot team can be employed for continuous coverage of a dynamic environment. In this paper, we propose a novel approach for creating multi-robot patrolling policies, which is fault-tolerant and self-adaptive. A dynamic priority queue and time-out replanning mechanism are maintained by each robot to schedule the tasks fault-tolerantly in the context of the market-based method. Hoplites framework is adapted by introducing a self-adaptive threshold adjustment and sharing mechanism to provide a high-level coordination. This work is demonstrated by a multi-robot patrolling task implemented on Robot Operating System (ROS). A flexible tool, Stage, is leveraged to provide the simulated environment. The experimental results validate the effectiveness and availability.