Improved Positioning of City Buses

In order to build an efficient transportation system of the future, public transport needs to be an attractive alternative offering a high quality travel experience. Future digital information systems for public buses will enrich the travelers experience by delivering relevant content - accurately on the right time and place. In this master thesis, design and implementation of an improved bus positioning system is presented with the objective to meet these demands. The system is based on a bank of particle filters used to track a non-linear, non Gaussian state model of the bus. The available input signals are GPS, wheel speed, and door opening sensors. The system utilizes a GIS database containing the street network and a local traffic system database with spatial and temporal information of the buses. The implementation is done in python, powered by the GeoDjango framework and the PostgreSQL database, and has been tested in buses containing a commercial digital information system. The system improves the positioning during GPS failure and increases reliability by combining all disposable information and presenting an accurate and redundant estimate.