|
Abstract:
|
In studies of driver safety, high-frequency data collection allows the study of complex patterns of driving. For example, Boer (2000) and Dawson et al (2006) have used a measure called entropy to quantify lateral vehicular control. To investigate such approaches, specific models are needed to capture key aspects of driving data. We propose a model that accommodates the inertia of the vehicle path, correlated errors, and the environment wherein vehicles are used. We compare simulated data based on this model versus actual data from elderly subjects, showing which model settings best fit empirical data. We also describe random-effects extensions of this model to capture the between-person variability in driving performance. The use of these models may aid in assessing driver safety, and in the development of onboard systems to warn at-risk drivers of possible performance impairments.
|
