Abstract
Advanced Driver Assistance Systems (ADAS) are aiming at enhancing the capabilities of the driver. The state of the driver is a critical parameter in decision making support algorithms, especially in systems that seize control either completely or partially from the driver. The driver state can be estimated by matching the driver performance with a driver model. When the model parameters are determined during practical driving conditions, the variation of these parameters may be used to interpret the driver state, and to use it to make ADAS more effective. This paper presents an approach how to derive driver model parameters (gains, delays, preview distance), based on a path tracking driver model. These parameters are estimated from the vehicle handling parameters during normal driving on a public road. First, a simulation research is carried out to find out whether the approach is able to distinguish between different driver attitudes. For this purpose, a new driver model is introduced, which allows varying speed, preview time and steering gain, while manoeuvring along arbitrary cornering and lane change conditions. Next, the method is applied on real driving situations, where the resulting model parameter variation (especially the preview time) is compared to workload measure SRR (Steering Reversal Rate). Different preview time results are obtained for drivers with different experience. Lower preview time, and therefore higher steering gain, appears to correspond to higher SRR and large steering rate variation.
KEYWORDS – driver state, driver model, vehicle handling, steering gain, preview time