Communication Jamming-Aware Robot Path Adaptation

Abstract

Robot networks are necessary for a variety of monitoring and surveillance applications where robots must communicate observations. This work considers robot networks that must operate in an area affected by potential communication jamming and excessive radio frequency noise. Specifically, we look at deploying a reinforcement robot to strengthen an existing network of robots that are already performing a monitoring task. The robot's objective is to find a location that maximizes the network's signal-to-noise ratio. We present a Gaussian process approach to predict areas of excessive radio frequency noise, enabling the robot to avoid these areas and relay more data. We evaluate our algorithm's performance in physical robot experiments using a Clearpath Jackal robot and a USRP software-defined radio to broadcast a jamming signal. Our results show major improvements in networking metrics over a baseline and demonstrate the need for jamming-aware robot planning.