Development of a GNSS-based multi-sensor vehicle navigation system.

摘要

A vehicle navigation system is developed that uses a combination of GPS positioning and dead reckoning sensors. A survey of dead reckoning sensors is presented, including examples and comparisons of different technologies. The dead reckoning sensors chosen for the final implementation were a low cost piezoelectric vibrating gyro and differential odometry provided by a vehicle's anti-lock braking system. Filtering of the data is done by a centralized Kalman filter with independent calibration filters for the dead reckoning sensors. Augmentations of GPS positioning are also investigated, focusing on clock coasting using an OCXO and height aiding. Several series of tests were performed to analyze the system's performance, culminating with tests in downtown Calgary under urban canyon masking and multipath conditions. Results from the testing indicate that an accuracy on the order of 10 to 20 metres can be achieved under most circumstances, but more intelligent algorithms or map matching are needed to control the large deviations which can be caused by poor GPS solutions.