DOA Estimation for Autonomous Systems in Complex Propagation Environments
Gunjan Verma, Fikadu T. Dagefu, Brian M. Sadler, Jeffrey N. Twigg, Jonathan Fink · 2018
We consider the fundamental problem of direction of arrival (DoA) estimation for autonomous agents in obstacle-rich, near-ground channels. Classical phase-based approaches to DoA estimation rely on the impingement of the direct path signal on an antenna array; when line of sight is rare or unobtainable such methods may yield meaningless estimates. In this paper, we present a Bayesian algorithm for DoA estimation based on the received signal strength (RSS) spatial gradient, where RSS measurements are collected while maneuvering in the environment. We develop a Bayesian methodology that incorporates smoothing and is robust to noise and outliers and provides uncertainty quantification. Experiments in both microwave and low VHF bands demonstrate the approach and show how useful DoA information can be obtained in dense cluttered environments. The results can be coupled with autonomous control, and are useful for multi-aaent coordination” source seekina, and localization.