Vehicle road departure detection using anomalies in dynamics


This research investigates the viability of detecting vehicle road departure via the measurements of anomalies in vehicle dynamics, especially under conditions when left and right tires experience imbalance of forces (split-μ condition). This approach is based on established low-order vehicle models to facilitate real-time implementation. Vehicle states are obtained from an INS system to obtain real-time estimates of model agreement with measured data obtained from a steer-by-wire experimental test vehicle (P1). Experimental maneuvers were conducted on various surface conditions, including four tires on dry asphalt, the two passenger tires on a low-friction patch, and all four tires on a low-friction surface. These vehicle states and steering moments were recorded from INS systems and torque transducers, in an effort to identify normal and abnormal driving modes. Results at high speeds show a yawrate mismatch between model and experimental measurements that gives a large enough signal-to-noise ratio to allow detection using anomalies in dynamics, whereas at lower speeds measurements of steering torque may be additionally needed to improve the signal-to-noise ratio. A combination of steering torque and yawrate measurements is needed to obtain comprehensive information to allow detection and determination of road departure events.

DOI: 10.1109/ACC.2016.7526662

15 Figures and Tables

Cite this paper

@inproceedings{Yang2016VehicleRD, title={Vehicle road departure detection using anomalies in dynamics}, author={Hang Yang and Derek McBlane and Christina Boyd and Craig Beal and Sean Brennan}, booktitle={ACC}, year={2016} }