This paper discussed the development of a unified terrain-aided framework for achieving minimal infrastructure localization of high speed vehicles operating in unstructured and harsh environments. The first step in the achievement of this goal was the development of a map-based ICP-EKF (Iterative Closest Point - Extended Kalman Filter) localization algorithm utilizing measurements from a scanning laser rangefinder in combination with dead-reckoning sensors. The next step was the identification of shortcomings of the ICP-EKF algorithm and the development of an entropy-based landmark augmentation metric to overcome the deficiencies. Finally, the paper detailed the integration of the information metric, the CSS (Curvature Scale Space) and the ICP-EKF algorithms to arrive at a unified localization framework. The developed localization framework has the ability to use measurements from both artificial and natural landmarks as and when they become available and was shown to be sufficiently generic to be used on a variety of autonomous land vehicles, by its application to a 4WD vehicle and an LHD (Load-Haul-Dump) truck. The results demonstrated the reliability and robustness of the proposed framework.
Terrain aided localization of autonomous vehicles
Geländeunterstützte Lokalisierung von autonomen Fahrzeugen
2002
8 Seiten, 7 Bilder, 8 Quellen
Aufsatz (Konferenz)
Englisch
Terrain Aided Localization of Autonomous Vehicles
NTIS | 2002
|Mid-water current aided localization for autonomous underwater vehicles
British Library Online Contents | 2016
|Extended terrain reconstruction for autonomous vehicles
SPIE | 2000
|Terrain Navigation Concepts for Autonomous Vehicles
NTIS | 1984
|