Automated Extraction of Driver Behaviour Primitives Using Bayesian Agglomerative Sequence Segmentation

The low-level building blocks of driver behaviour have been shown to exhibit statistical patterns such as periods of turning, braking and acceleration, as well as different combinations of these. Collectively, these patterns can be regarded as a language of "driving primitives." This allows us to reason about more meaningful driving maneuvers, e.g. overtaking, parking, by treating them as sequences of primitives. In this paper we introduce a method for automatically finding the boundaries between primitives, which is important when analysing large volumes of raw sensor data that can be generated in ITS applications. Our method is cost-effective, completely unsupervised and requires minimal preprocessing. We demonstrate the potential of our approach via an experiment with genuine data from an inertial measurement unit.