Optimizing the area coverage of networked UAVs using multi-agent reinforcement learning

Abstract

Wireless sensor networks (WSNs) have been widely used in various area coverage applications which require the monitoring and surveillance of systems with spatiotemporally varying variables or parameters. One important task in the implementation of WSNs for area coverage and monitoring purposes is the determination of the solution for the optimal coverage problem. This paper describes that the formulation of the area coverage problem can be modeled using Markov game modeling formalism whereas the optimal joint state-action policy for each agent which also takes into consideration the group objective can be computed using multi-agent Q-learning iterative processes using multi-agent reinforcement learning framework. Simulation results are presented to illustrate the proposed iterative learningbased area coverage solution approach.