Estimation of Driver's Risk Awareness from Physiological Data for Smart Assistance Systems

Abstract

utomated and semi-automated maneuvering will have to be monitored by the driver. To fulfil this task, the driver needs to have adequate risk awareness, which should represent the actual risk of the maneuver within the traffic situation. This work aims at deriving a model of driver’s risk awareness from physiological measurements and traffic context, which were acquired from simulator experiments. Given these prerequisites, a Bayesian Network may be postulated and trained with the data to estimate the risk awareness as internal driver’s state. The results of our study show the general feasibility of our approach. Future assistance systems may take this information and adapt to the user’s risk awareness accordingly.