Particle Filtering for Location Estimation
Thesis DisciplineElectrical Engineering
Degree GrantorUniversity of Canterbury
Degree NameMaster of Engineering
Vehicle location and tracking has a variety of commercial applications and none of the techniques currently used can provide accurate results in all situations. This thesis details a preliminary investigation into a new location estimation method which uses optical environmental data, gathered by the vehicle during motion, to locate and track vehicle positions by comparing said data to pre-recorded optical maps of the intended location space. The design and implementation of an optical data recorder is presented. The map creation process is detailed and the location algorithm, based on a particle filter, is described in full. System tests were performed offline on a desktop PC using real world data collected by the data recorder and their results are presented. These tests show good performance for the system tracking the vehicle once its approximate location is determined. However locating a vehicle from scratch appears to be infeasible in a realistically large location space.