Autonomous navigation to an RF source in multipath environments

Abstract

We consider the problem of autonomous navigation of a mobile robot to a radio-frequency (RF) source in a sparse multipath environment, in which the channel has a few dominant propagation paths. We consider a scenario in which the mobile robot is equipped with a moderately directional antenna but has no knowledge of the channel path-loss exponent or number or position of the multipath reflectors in the environment. The directional antenna allows the mobile robot to use angle-of-arrival (AoA) information to infer the direction in which RF power is arriving from one or more propagation paths; however, the presence of multipath and the wide beam of the directional antenna makes it very challenging to determine the location of the source from measurements at a single location. To overcome this, the mobile robot uses measurements at multiple locations and fuses the measurements using a particle filter. In turn, the robot is moved in the direction of one of the surviving particles after each particle filter update. We evaluate the performance experimentally using software-defined radios and a Turtlebot 2 robot in a controlled multipath environment, and we also simulate the performance using a ray-tracing model. Both sets of results show that this is a viable approach for navigating an autonomous vehicle to an RF source.