Abstract
Developing a driving state estimation framework is important to improve the core technologies of individual adaptation technique. This paper proposes a scientific method to detect deviated driving by using individual reference driver behavior model. First, the model treats the car-following state as a spring-mass-damper system. Next, to diagnose the driver, the deviated driving is detected from the distribution of deviation of each state variable compared to the reference model. The driver model parameters are identified at every certain sampling time duration for the diagnosis function. Finally, the proposed algorithm is implemented in the test vehicle to examine its feasibility.
