Research On Augmented Positioning Technique In Car Navigation System

GPS positioning is with high absolute accuracy, but the reception can be blocked by obstacles between GPS receiver and satellites, such that positioning fails. Dead Reckoning (DR) can output accurate positioning during short term, but the positioning error tends to grow unboundedly. Map Matching (MM) is an aided positioning technique based on digital map. Fusing the Information of GPS, DR and MM can improve the positioning accuracy, reliability, and applicability of positioning system, which is a key technique in multi-sensors positioning system. This research is a part of the development of low cost car navigation system, which aims to develop a real-time, moderate accuracy and robust multi-sensors positioning method applicable to low cost navigation products.Firstly, the Integrated Kalman Filter (KF) and modified Federal Kalman Filter (FKF) for GPS/DR system are proposed and compared. Since FKF is more fault-tolerant and its design is flexible, FKF is selected as the basic fusion algorithm in GPS/DR system. In the implementation of FKF, two sets of distribution factors used for reset of subsystems'states and covariance matrix are proposed according to the difference of estimation accuracy between position states and velocity states, acceleration states. One set is for position states, the other set is for velocity and acceleration states. Distribution factors are adapted according to the GPS's positioning error. In order to reduce the computation load in main filter's fusion process, the diagonal matrix of state covariance is substituted for state covariance matrix, under the assumption that the correlativity between states'noise is low. Simulations have verified that the fusion accuracy is decreased slightly, while computation load is reduced markedly, thus the real-time performance is improved. Innovations of FKF are partly fed back to DR equations to restrain the accumulation of error according to the characteristic of innovations. Real-time running on road has proved that the proposed FKF is with the advantage of real-time, high fusion accuracy and reliable operation.The FKF's performance is heavily dependent on the accuracy of plant model, its computation load is relative heavy, and further more it is with the problem of numeric instability. Such disadvantages exclude EKF's application in low computation power processor. An algorithm based on fuzzy logic is proposed for GPS/DR system. Three inputs for fuzzy logic are estimation of GPS's error, estimation of DR's error and...