Evaluation of driving skills using an HMM-based distance measure

In this paper, we address a model-based objective measure for the evaluation of driving skills between different drivers. This metric is based on a stochastic distance between a pair of hidden Markov models (HMMs) each of which is trained for an individual driver. The emphasis of comparison is on the differences between the stochastic somatosensory processes of human driving skills. To evaluate the adequacy of the metric, we developed a driving simulator and carried out an experiment that collected the driving data of many novice drivers and an expert driver. The objective measures were computed between each novice driver and the expert driver, and they were compared with the subjective judgement of each novice driver´s skills made by the expert driver. Analysis results showed high agreement between the two measures, supporting that the objective metric is a suitable descriptor for the differences in driving skills. The findings of this work can contribute to developing a driving simulator for training with an objective assessment function of driving skills.