Modeling of Driving Behavior When Approaching an Intersection Based on Measured Behavioral Data on An Actual Road

Abstract

It is important to develop a driving assistance system that conforms to individual driving behavior. An individual's usual driving behavior is generally adapted to road and traffic situation; therefore, we propose that assistance be provided when that behavior deviates from normal. We developed specially equipped vehicles to collect natural behavioral data from repeated driving experiments on an actual road. We subsequently developed a Bayesian network model of occurrence of behavioral events from the data obtained when approaching a specific intersection with a stop sign. Any deviation from usual operation behavior can be estimated using this model. Unusual behavior can be regarded as inappropriate behavior and therefore its detection can be applied to driving assistance methods, such as warning presentation.