Using High-Definition maps for precise urban vehicle localization

Abstract

Global Navigation Satellite Systems (GNSS) based localization in the context of Intelligent Transportation Systems, Advanced Driver Assistance Systems and ultimately the development of autonomous vehicles often requires accuracy with integrity. Low-cost data fusion algorithms for positioning that integrate GNSS and additional in-vehicle sensor information are able to handle short term GNSS outages and can provide increased availability and accuracy. Challenging are non-line-of-sight effects that are often present in urban areas. As those can hardly be modeled accurately, an unobservable position bias is often introduced which violates the estimated confidence. Therefor, mass-market applicable localization algorithms that mitigate their influence are desired to allow GNSS-based positioning for safety critical applications. With the beginning market introduction of high-definition maps (HD maps) by major map providers, another inexpensive tool to enhance data fusion based localization algorithms is available. Those maps offer means to further improve the accuracy and integrity of vehicle position estimation, especially in challenging urban environments. In this paper, a particle filter based localization algorithm is presented that integrates HD maps, GNSS and vehicle odometry measurements to demonstrate the benefits of HD maps using real world measurements.