An energy-efficient method for multi-robot reconnaissance in an unknown environment

Abstract

Autonomous robots have significant potential for reconnaissance and environmental monitoring applications. Ground robots, in particular, are performing reconnaissance missions in places that are too hazardous for humans. However, these robots are constrained by energy limitations that are impacted by uncertain environments and harsh terrains. The purpose of this work is to develop methods for improving the efficiency of reconnaissance missions through energy awareness. To address such limitations, robot energy usage is spatially modeled with a Gaussian Process (GP) through measurements collected during the mission. The resulting energy predictions are incorporated into a centralized waypoint-based optimization with the goal of minimizing the uncertainty of a spatio-temporal field, subject to ensuring the robots' return to their respective starting locations for refueling. Simulation results for a 3-robot system demonstrate the effectiveness of incorporating energy predictions into reconnaissance missions.