Concept, Implementation, and Performance Comparison of a Particle Filter for Accurate Vehicle Localization Using Road Profile Data

Abstract

A precise knowledge of the road profile ahead of the vehicle is required to successfully engage a proactive suspension control system. If this profile information is generated by preceding vehicles and stored on a server, the challenge that arises is to accurately determine one’s own position on the server profile. This article presents a localization method based on a particle filter that uses the profile observed by the vehicle to generate an estimated longitudinal position relative to the reference profile on the server. We tested the proposed algorithm on a quarter vehicle test rig using real sensor data and different road profiles originating from various types of roads. In these tests, a mean absolute position error of around 1 cm could be achieved. In addition, the algorithm proved to be robust against local disturbances, added noise, and inaccurate vehicle speed measurements. We also compared the particle filter with a correlation-based method and found it to be advantageous. Even though the intended application lies in the context of proactive suspension control, other use cases with precise localization requirements such as self-driving cars might also benefit from our method.