Estimation and prediction of vehicle dynamics states based on fusion of OpenStreetMap and vehicle dynamics models

Jiang, Kun; Victorino, Alessandro; Charara, Ali · 2016 · Crossref

DOI: 10.1109/ivs.2016.7535387

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of estimating and predicting vehicle dynamics states to enhance active safety systems. While existing systems like ABS and ESP react to events that have already occurred, the authors argue that predicting potential dangers before they manifest allows for earlier corrective action, particularly at high speeds. The core problem is that critical safety parameters, such as tire-road contact forces and vehicle side slip angle, are difficult to measure with low-cost sensors. Furthermore, traditional observers estimate current states but lack the ability to predict future dynamics. To address this, the study proposes a novel approach that fuses data from inertial sensors, GPS, and OpenStreetMap (OSM) to estimate current states and predict future risks based on upcoming road geometry. The methodology involves processing OSM data to create a topological representation of the road using "Critical Points" (CPs) and connecting "corridors." CPs are sparse locations where dynamics change significantly, such as roundabouts or slippery regions, attributed with properties like curvature, friction coefficient, and bank angle. Corridors between CPs are modeled as straight lines or clothoids. Vehicle localization is achieved via an Extended Kalman Filter (EKF) combining GPS and inertial measurements. The system uses road geometry from OSM to provide redundant information for current state estimation and to predict future dynamics by assuming the vehicle follows the planned path. Safety is evaluated using three risk assessment indexes: lateral load transfer ratio, lateral skid ratio, and stopping distance. The estimation employs a linear tire model and bicycle model for sideslip angle, and a double-track model with Dugoff tire model for individual tire forces. Experimental validation was conducted using the DYNA instrumented vehicle, which features direct measurement of vertical and lateral tire forces to serve as ground truth. The test trajectory included 53 critical points and 52 corridors, with analysis focusing on a segment involving successive turns at an average speed of 50 km/h. Results demonstrated that inertial sensor-based estimation accurately followed vertical force variations but struggled with sideslip angle estimation when model parameters were inaccurate. Conversely, the OSM-based method provided accurate sideslip angle estimates by deriving correct slip stiffness from map data but failed during unplanned maneuvers like lane changes. The fused approach, combining both methods via EKF, yielded superior estimation accuracy for vertical forces, lateral forces, and sideslip angles compared to either method alone. The significance of this work lies in demonstrating that integrating digital map data with vehicle dynamics models can improve the accuracy of state estimation and enable the prediction of future safety risks. By leveraging OSM to provide information on road geometry and friction, the system can assess potential dangers up to 300 meters ahead. This approach offers a pathway for more proactive safety systems that can warn drivers or control vehicle stabilization before hazardous conditions are encountered, addressing the limitations of reactive safety technologies.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• situational awareness