A moving fixed-interval filter/smoother for estimation of vehicle position using odometer and map-matched GPS

TL;DR

This paper develops optimal real-time and post-processing estimators for vehicle position using odometer and GPS data, demonstrating improved accuracy over classical methods through simulations and real data tests.

Contribution

Introduces a set of optimal estimators based on a simple error model, including asymptotically minimum variance unbiased estimator and flexible fixed interval filters and smoothers.

Findings

Estimators outperform classical Kalman filter in accuracy.

Good performance demonstrated with simulated and real data.

Flexible filters and smoothers adapt to different estimation needs.

Abstract

This paper presents some optimal real-time and post-processing estimators of vehicle position using odometer and map-matched GPS measurements. These estimators were based on a simple statistical error model of the odometer and the GPS which makes the model generalizable to other applications. Firstly, an asymptotically minimum variance unbiased estimator and two optimal moving fixed interval filters which are more flexibles are exposed. Then, the post-processing case leads to the construction of two moving fixed interval smoothers. These estimators are tested and compared with the classical Kalman filter with simulated and real data, and the results show a good accuracy of each of them.