Cooperative Position and Orientation Estimation with Multi-Mode Antennas

Kurzfassung

Robotic multi-agent systems are envisioned for planetary exploration, but also for terrestrial applications like search and rescue and environmental monitoring. Autonomous operation of robots requires estimations of their positions and orientations, which are obtained from the direction-of-arrival (DoA) and the time-of-arrival (ToA) of radio signals. For cooperative radio localization, signals are exchanged among all agents. Within this thesis, we estimate the signal DoA and ToA using a multi-mode antenna (MMA). An MMA is a single antenna element, where multiple orthogonal current modes are excited by different antenna ports. So far, MMAs have been considered for multiple-input multiple-output (MIMO) communications. This thesis provides a first study on the use of MMAs for cooperative position and orientation estimation. We specifically explore the DoA estimation capabilities of MMAs. Assuming the agents of a cooperative network are equipped with MMAs, lower bounds on the achievable position and orientation accuracy are derived. We realize a gap between the theoretical lower bounds and real-world performance of a cooperative radio localization system. The reason are biased estimates due to deviations of antenna and transceiver calibration parameters. Consequentially, we theoretically analyze in-situ antenna calibration. We further introduce an algorithm for in-situ calibration of arbitrary multiport antennas, including MMAs, and show its effectiveness by simulation. To also improve calibration during operation, we propose cooperative simultaneous localization and calibration (SLAC), which leverages the large number of observations within a cooperative network. We show that cooperative SLAC is able to estimate antenna responses and ranging biases of the agents together with their positions and orientations, without external sensors. For that, a Bayesian filtering algorithm for cooperative SLAC is derived and simulations are performed, showing considerable improvements of position and orientation accuracy compared to mere localization. Finally, we validate the results from theory and simulation by experiments with four robotic rovers equipped with software-defined radios (SDRs). On one rover, a four-port MMA is installed. We experimentally demonstrate DoA estimation with a single MMA, in-situ calibration of an MMA and cooperative SLAC. In conclusion, we show that DoA estimation with an MMA is feasible, and accuracy can be improved by in-situ calibration and SLAC.