| dc.description.abstract | This thesis summarizes the development of a collaborative localization algorithm simulation environment and the implementation of collaborative localization using Ultra-Wideband ranging in autonomous vehicles. In the developed simulation environment, multi-vehicle scenarios are testable with various sensor combinations and configurations. The simulation emulates the networking required for collaborative localization and serves as a platform for evaluating algorithm performance using Monte Carlo analysis. Monte-Carlo simulations were run using a number of situations and vehicles to test the efficacy of UWB sensors in decentralized collaborative localization as well as landmark measurements within an extended Kalman filter. Improvements from adding Ultra-Wideband ranging were shown in all simulated environments, with landmarks offering additional improvements to collaborative localization, and with the most significant accuracy improvements seen in GNSS-denied environments. Physical experiments were run using a by-wire GEM e6 from Autonomous Stuff in an urban environment in both collaborative and landmark setups. Due to higher than expected INS certainty, adding UWB measurements showed smaller improvements than simulations. Improvements of 9.2 to 12.1% were shown through the introduction of Ultra-Wideband ranging measurements in a decentralized collaborative localization algorithm. Improvements of 30.6 to 83.3% were shown in using UWB ranging measurements to landmarks in an Extended Kalman Filter for street crossing and tunnel environments respectively. These results are similar to the simulated data, and are promising in showing the efficacy of adding UWB ranging sensors to cars for collaborative and landmark localization, especially in GNSS-denied environments. In the future, additional moving vehicles with additional tags will be tested and further evaluations of the UWB ranging modules will be performed. | en |
