Vehicle Localization Using a Traffic Sign Map

Jos Elfring, Subodh Dani, Siavash Shakeri, Hamed Mesri, Jan Willem van den Brand

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

Abstract

High-definition (HD) map data is considered a pre-requisite for automated driving since it delivers a rich source of information insensitive to line-of-sight limitations and weather conditions. In order to exploit HD map data, a vehicle must be able to localize within the map. Vehicle localization solutions based on only GNSS data, odometry and IMU data are insufficient both in terms of accuracy and reliability. Cameras are used in production vehicles and therefore offer a realistic source of additional sensor information that could be exploited for improving localization accuracy and robustness.This work proposes a vehicle localization solution that exploits a traffic sign map and images from a single camera for improving localization accuracy. A particle filter is used for estimating the vehicle's position. Likelihoods are computed in the image domain to avoid the need for estimating 3D positions of 2D detections in camera images. During an experimental evaluation, the localization error is shown to decrease to approximately 0.7 m in both lateral and longitudinal direction.

Original languageEnglish
Title of host publication2020 IEEE 23rd International Conference on Intelligent Transportation Systems, ITSC 2020
PublisherInstitute of Electrical and Electronics Engineers
Number of pages6
ISBN (Electronic)978-1-7281-4149-7
DOIs
Publication statusPublished - 20 Sep 2020
Event23rd IEEE International Conference on Intelligent Transportation Systems, ITSC 2020 - Rhodes, Greece
Duration: 20 Sep 202023 Sep 2020
https://www.ieee-itsc2020.org/

Conference

Conference23rd IEEE International Conference on Intelligent Transportation Systems, ITSC 2020
Abbreviated titleITSC2020
Country/TerritoryGreece
CityRhodes
Period20/09/2023/09/20
Internet address

Fingerprint

Dive into the research topics of 'Vehicle Localization Using a Traffic Sign Map'. Together they form a unique fingerprint.

Cite this