Improved positioning of motor vehicles through secondary information sources

Publication Type:
Issue Date:
Full metadata record
Files in This Item:
Filename Description Size
Thumbnail01front.pdf607.11 kB
Adobe PDF
Thumbnail02whole.pdf16.51 MB
Adobe PDF
NO FULL TEXT AVAILABLE. This thesis contains 3rd party copyright material. ----- Positioning systems are a key enabling technology for many Intelligent Transport Systems applications. Route guidance, fleet management, stolen vehicle recovery, emergency assistance, and location specific traveller information systems are a few examples of applications requiring accurate position measurement of motor vehicles. Positioning systems, especially those based on the measurement of wave propagation properties, are not, in general, specifically designed to track motor vehicles. Restrictions imposed by the road network, dynamic limitations of the vehicle, road rules, driver behaviour, and other behaviours specific to motor vehicles travelling in an urban environment are not incorporated into the position estimation process; valuable sources of information are ignored. Although these extra sources of information do not specifically identify where the vehicle is located, they can augment the primary position measurements. This thesis identifies, classifies, and models supplementary information sources concerning the position of motor vehicles. Algorithms are developed to incorporate the various supplementary information sources into the positioning process. A hierarchical positioning systems architecture is proposed to provide a framework in which all of the algorithms can be seamlessly integrated. The majority of the research presented concentrates on two secondary information sources; the map of the road network and the vehicle dynamics. A maximum à posteriori (MAP) estimator is presented which uses the road map information to translate the primary position and velocity measurements onto the road network; a process referred to as map-aided estimation. Vehicle dynamics are introduced via a novel Kalman filter, referred to as the Spatially Reduced Kalman Filter (SRKF), which is matched to the reduced dimensionality of the map aided estimates. The map-aided estimation/SRKF combination should remove more measurement noise than the more traditional approach of a decoupled two-dimensional Kalman Filter and the SRKF's model of the vehicle's dynamics should be more accurate. Position and velocity ambiguities can arise during the map-aided and SRFK estimation processes. These ambiguities are resolved using track splitting algorithms, adapted from target tracking techniques, which determine which road the vehicle is currently travelling on. The track splitting approach involves the generation and maintenance of a series of hypotheses representing the possible time/space trajectories that the vehicle could have followed given the observed measurements. The best hypothesis determines the current position and velocity for the vehicle. To augment the decision making processes involved in evaluating the hypotheses, tertiary information sources, such as road rules and driver behaviour, are examined. By integrating extra information sources into the motor vehicle positioning process the accuracy of motor vehicle positioning is improved. In addition the reported positions have increased meaning and utility as they are also referenced to local features and not just a two-dimensional coordinate. The comprehensive and novel approach to motor vehicle positioning presented in this thesis has a strong theoretical basis and the open hierarchical architecture proposed readily facilitates the incorporation of any further information sources should they be identified and become available.
Please use this identifier to cite or link to this item: