Real-time identification method of driver model with steering manipulation

Abstract

This study proposes a method for real-time identification of a driver model. The proposed method requires only the yaw rate sensor, the steering angle sensor, and velocity sensors that are usually installed in the production car. The identification algorithm involves the division of the recorded data, prefiltering of the divided data, estimation of the driver's desired response, and identification. The prefilter extracts the driver's involuntary response that can be modelled in a simple form. The ideal car response that the driver attempts to track is estimated from the recorded data, and this response is provided to the identification algorithm of the feedback driver model for error tracking. These newly developed methods enable real-time identification under actual driving conditions. The driving simulator experiments and the actual driving tests were performed, and the proposed method was validated. The results show that the time history of the variation in the driver's characteristics can be realized in real time using the proposed method.

Drowsiness estimation from identified driver model

“…Firstly, the real-time identification method is briefly described (Tokutake et al, 2013). A driver's actions are based on the sensing of many kinds of information, including information about the road surface condition, curvature of the road, and vehicle behaviour.…”

“…The output of the driver model is the steering history. In the previous study (Tokutake et al, 2013), a driver's involuntary responses to the yaw dynamics of the car were modelled as follows.δ…”

“…A low-cost and reliable drowsiness detection system is necessary for practical use. In contrast, we proposed a real-time identification method for a driver model (Tokutake, Sugimoto, & Shirakata, 2013). In this method, the driver's involuntary steering response to the car's motion is formulated as a differential equation using the measured rate-gyro and steering-angle data.…”

“…In our study, a driver model was identified and subjected to drowsiness detection tests. The real-time identification method was established by Tokutake et al (2013). In the next step, an estimation method for the driver's drowsiness using the identified driver model was constructed.…”

“…Then the measured drowsiness and the identified driver model of the driving test were compared. Here, the identification process is based on the real-time method that was reported in Tokutake et al (2013). Finally, the estimation function for drowsiness was constructed using the identified parameters and subsequently validated.…”