Driving course prediction for vehicle handling maneuvers


This paper aims at predicting the future driving course, which we define as a combination of two bifurcating channels - future speed and steering action that in turn derive a future driving trajectory during a curve. In defining the relation of these two channels, human factors, such as the stressfulness, comfort level, and skillfulness of the driver, are paid particular attention to. While the modeling and forecast of speed and steering angle are to some extent separated, a hidden Markov model (HMM) that's designed to mimic driver's intention integrates them by making subjective corrections. The proposed algorithm has been proved effective on realistic driving data based on a prototype vehicle at Ford.

In 2012 American Control Conference (ACC).
@inproceedings{liu2012driving, title={Driving course prediction for vehicle handling maneuvers}, author={Liu, Ruoqian and Yu, Hai and McGee, Ryan and Murphey, Yi L}, booktitle={2012 American Control Conference (ACC)}, pages={2096–2101}, year={2012}, organization={IEEE}}