Behavioral Entropy as an Index of Workload

Abstract

Attentive well-trained drivers exhibit smooth control behavior with minimal discontinuities under non-taxing driving conditions. As drivers engage in activities different from the low-level driving tasks of lane keeping and car following, discontinuities enter into their control behavior profiles in the form of error corrections or startled responses. These corrective actions are responses to driver's own assessment of the state of the vehicle after attention is redirected back to the primitive' task. These smoothness-reducing discontinuities result in less predictable control behavior. To quantify the effect of drivers' responses to unacceptable vehicle states that build up while not fully attending, an entropy measure is used as an index of behavioral predictability and used to characterize the workload induced by different tasks. Workload is considered high over a given time interval when behavioral entropy is high. Based on data obtained in the Nissan-CBR driving simulator, the entropy measure is demonstrated to clearly and significantly quantify the effect of several tasks.