Similar to speech where phonemes form words, it has been established that the smallest meaningful units of a driving pattern are termed “drivemes”; drivemes form maneuvers, and maneuver sequences form a navigation route. Therefore, analyzing how these drivemes and maneuvers are executed by drivers can provide information on how the driver controls the vehicle and how driving performance varies over time.
HMM (Hidden Markov Models) is an effective statistical tool to model driver behavior for the following reasons:
- HMM can model the stochastic nature of the driving behavior, providing sufficient statistical smoothing while offering effective temporal modeling.
- The variations in driving signals across the drivers can be modeled (driver identification) or suppressed (driver-independent route models) according to the requirements of the desired task.