- مبلغ: ۸۶,۰۰۰ تومان
- مبلغ: ۹۱,۰۰۰ تومان
Vehicular positioning with GPS/IMU has been studied a lot to increase positioning accuracy. The positioning algorithms mainly use DR (Dead Reckoning) which uses EKF (Extended Kalman Filter). It is basic and very important core technology in positioning section. However, EKF has a major drawback in that it is impossible to make very accurate system and measurement models for a real environment. In this work, we propose an algorithm to estimate vehicle’s position as distribution form, and to control the system and measurement noise covariance to compensate for this major disadvantage. The proposed method to control noise covariance is independently processed, using fading factor and sensor error while considering the driving condition.
In this paper, we proposed the method controlling EKF filter noise covariance which consisted of system and measurement noise covariance. They are independently, automatically adjusted by many factors. A system noise covariance Q indirectly influences the performance of system model in EKF by using Lamda λ. A measurement noise covariance R is adjusted by the driving conditions: the driving environments, the driving state. The result of positioning by using the proposed method is more reliable and accurate than only using EKF. In special situation, the performance of the proposed method is even better than the expensive instrument using high-cost RTK. In low-multipath area where RTK has quite good performance, the performance of the proposed algorithm is around 0.4 m better than one using only EKF.