An investigation of goal assignment for a heterogeneous robotic team to enable resilient disaster-site exploration

Abstract

In this work we present a framework and quantitative metric for comparing the performance and resiliency of goal assignment strategies in the context of autonomous, disaster-site exploration using a heterogeneous team of robots. We propose three feasible policies for assigning user-defined Areas-of-Interest amongst a team of an Unmanned Ground Vehicle (UGV) and a squadron of Unmanned Aerial Vehicles (UAVs). We characterize the performance of these strategies in a series of simulations using our framework and take into consideration the real-world constraints of stochastic UAV flight time failures, stochastic communications failure, and limited UAV battery life. Finally, we test the sensitivity of the proposed assignment methods to varying levels of UAV flight failure, poor communications, and increased exploration duration.