Waypoint-Informed Localization (WiL): A Novel Integration of 2D Maps and Onboard Sensors for Robust GNSS-Denied Navigation

Abstract Reliable vehicle localization in global navigation satellite system (GNSS)-denied urban environments remains challenging owing to inertial drift during extended outages. This paper introduces waypoint-informed localization (WiL), a framework that leverages digital map information and onboard sensors to maintain navigational accuracy. The system employs the onboard diagnostic II (OBD-II) velocity integration framework (OVIF), which combines OBD-II vehicle speed with estimates from a semantic-aware stereo vision-based navigation system. During GNSS outages, WiL derives position and azimuth pseudomeasurements by integrating a vehicle kinematic model with two-dimensional road geometries, specifically road centerlines and edges. These estimates serve as aiding measurements for an extended Kalman filter, thereby bounding errors of a three-dimensional reduced inertial sensor system. Experimental validation demonstrates that the WiL-OVIF system achieves a 74.6% reduction in distance-traveled error compared with baseline methods, maintaining a localization error of 0.128% over a 2.8-km GNSS outage.