Vehicle guidance by using GPS aided Kalman filter

In this thesis, a novel approach that integrates global positioning systems (GPS) and Kalman filtering is presented. In order for a mobile robot to localize and navigate itself, a novel measurement process is developed which makes use of two low-cost GPS units. Using these ordinary, low-cost GPS receivers, a completely differential GPS-like system is achieved algorithmically in the absence of real DGPS service. One of these units is placed on an accurately surveyed geographical point and it is taken as a reference station for the other mobile unit. They are presumed to see the same satellites, hence the adoption of same measurement noise characteristics is considered to be appropriate. After the measurements are differentially corrected, a discrete Kalman filter algorithm is adopted to estimate optimally the position of a robot vehicle in order to navigate itself autonomously. The noise sequences of the Kalman filter are accepted as zero mean white Gaussian, and different filter performances are obtained by adjusting the parameters of the filter. In addition, trajectory estimation of the vehicle is realized using Kalman filter technique. Therefore, mapping of a specific geographical field in latitude and longitude is obtained. The results of the experimental testing of the DGPS algorithm and Kalman filtering show the effectiveness of the proposed approach.