Yulan Liang

How dangerous is looking away from the road? Algorithms predict crash risk from glance patterns in naturalistic driving.

Objective: In this study, the authors used algorithms to estimate driver distraction and predict crash and near-crash risk on the basis of driver glance behavior using the data set of the 100-Car Naturalistic Driving Study. Background: Driver distraction has been a leading cause of motor vehicle crashes, but the relationship between distractions and crash risk lacks detailed quantification. Method: The authors compared 24 algorithms that varied according to how they incorporated three potential contributors to distraction—glance duration, glance history, and glance location—on how well the algorithms predicted crash risk. Results: Distraction estimated from driver eye-glance patterns was positively associated with crash risk. The algorithms incorporating ongoing off-road glance duration predicted crash risk better than did the algorithms incorporating glance history. Augmenting glance duration with other elements of glance behavior—1.5th power of duration and duration weighted by glance location—produced similar prediction performance as glance duration alone. Conclusions: The distraction level estimated by the algorithms that include current glance duration provides the most sensitive indicator of crash risk. Application: The results inform the design of algorithms to monitor driver state that support real-time distraction mitigation systems.

Nonintrusive Detection of Driver Cognitive Distraction in Real Time Using Bayesian Networks

Driver distraction has become an important and growing safety concern as the use of in-vehicle information systems (IVISs), such as cell phones and navigation systems, continues to increase. One approach to allowing people to benefit from IVISs without compromising safety is to create adaptive IVISs that adjust their functions according to driver and roadway state. A critical element of adaptive IVISs involves monitoring driver distraction in real time; with such a monitoring function it is possible to mitigate that distraction. This study applied Bayesian networks (BNs), a data mining method, to develop a real-time approach to detecting cognitive distraction using drivers’ eye movements and driving performance. Data were collected in a simulator experiment involving 10 participants who interacted with an IVIS system while driving. BN models were trained and tested to investigate the influence of three model characteristics on distraction detection: time history of driver behavior, inclusion of hidden nodes in the model structure, and how data are summarized and the length of training sequences. Results showed that BNs could identify driver distraction reliably with an average accuracy of 80.1%. Dynamic BNs (DBNs) that consider time dependencies of driver behavior produced more sensitive models than static BNs (SBNs). Longer training sequences improved DBN model performance. Blink frequency and fixation measures were particularly indicative of distraction. These results demonstrate that BNs, especially DBNs, can detect driver cognitive distraction by extracting information from complex behavioral data. Potential applications include the design of adaptive in-vehicle systems and the evaluation of driver distraction.

A Looming Crisis: The Distribution of Off-Road Glance Duration in Moments Leading up to Crashes/Near-Crashes in Naturalistic Driving

Long glances away from the road present a significant risk to driving safety. The tail of the distribution for in-vehicle glance duration has been proposed to be more relevant to crash risk than the mean of the distribution. Using data collected in the 100-Car Naturalistic Driving Study (Dingus et al., 2006), this study examined the changes in the distribution of driver off-road glance duration, as well as in the frequency of such glances, at different time points before crash/near-crash events and compared with baseline (i.e., normal) driving. The results showed that the shape of the distribution of off-road glance duration at the onset of the precipitating factor of crashes/near-crashes was similar to the distribution in epochs during the preceding 25 seconds; nonetheless, the frequency of off-road glances increased in the approach to crashes/near-crash events. Compared with baseline epochs, drivers in crashes/near-crash events tended to look away from the forward road more often and with longer duration (i.e., with thicker tail end of the distributions; exceeding 1.7 seconds as observed in the distributions). It suggests that frequent off-road glances longer than 1.7 seconds present a high-risk glance pattern in the seconds preceding a safety- critical event and that the 2.0 second-threshold that is frequently cited in defining dangerously long off-road glances might be a liberal estimation.

Reading Text While Driving Understanding Drivers’ Strategic and Tactical Adaptation to Distraction

Objective In this study, we investigated how drivers adapt secondary-task initiation and time-sharing behavior when faced with fluctuating driving demands. Background Reading text while driving is particularly detrimental; however, in real-world driving, drivers actively decide when to perform the task. Method In a test track experiment, participants were free to decide when to read messages while driving along a straight road consisting of an area with increased driving demands (demand zone) followed by an area with low demands. A message was made available shortly before the vehicle entered the demand zone. We manipulated the type of driving demands (baseline, narrow lane, pace clock, combined), message format (no message, paragraph, parsed), and the distance from the demand zone when the message was available (near, far). Results In all conditions, drivers started reading messages (drivers’ first glance to the display) before entering or before leaving the demand zone but tended to wait longer when faced with increased driving demands. While reading messages, drivers looked more or less off road, depending on types of driving demands. Conclusions For task initiation, drivers avoid transitions from low to high demands; however, they are not discouraged when driving demands are already elevated. Drivers adjust time-sharing behavior according to driving demands while performing secondary tasks. Nonetheless, such adjustment may be less effective when total demands are high. Application This study helps us to understand a driver’s role as an active controller in the context of distracted driving and provides insights for developing distraction interventions.

Driver Cognitive Distraction Detection Using Eye Movements

Driver distraction is an important safety problem [518]. The results of a study that tracked 100 vehicles for one year indicated that nearly 80% of crashes and 65% of near-crashes involved some form of driver inattention within three seconds of the events. The most common forms of inattention included secondary tasks, driving-related inattention, fatigue, and combinations of these [325]. In-vehicle information systems (IVIS), such as navigation systems and internet services, introduce various secondary tasks into the driving environment that can increase crash risk [5,516]. Thus, in future it would be very beneficial if IVIS could monitor driver distraction so that the system could adapt and mitigate the distraction. To not disturb driving, non-intrusive and real-time monitoring of distraction is essential.

Reading while driving: a study on drivers' strategies of in-vehicle task initiation

In spite of increased public awareness and an ever-growing body of research, driver distraction remains an important safety concern. Reading text while driving may be especially detrimental by imposing both visual interference and extra cognitive demand on drivers. However, in most cases, drivers do not passively respond to such a task, they actively decide when to perform the task. The current study investigated drivers’ strategies with respect to the initiation of text reading when faced with fluctuations in driving demand. Text messages were made available to participants shortly before the vehicle entered an area with increased driving demands (demand zone). Participants were asked to read the message before the end of the trial, but were free to decide when and how to read the messages. We manipulated the type of driving demand (baseline; three levels of high demand—narrow lane, pace clock, and ultimate [narrow + pace clock]), format of text (no message; paragraph; parsed), and the distance from start of message to the demand zone (near; far). The results showed that in most cases drivers started to read the text message before or in the demand zone even though they would have enough time to complete the task after passing the area. Nonetheless, task initiation time (measured from the appearance of the message to the drivers’ first glance to the display) increased when driving was demanding compared to the baseline condition. It suggests that drivers may be sensitive to transitions from low to high demand when initiating secondary tasks, however are not discouraged from such activities when demands are already elevated.

Memory for a hazard is interrupted by performance of a secondary in-vehicle task

Driver visual distraction is known to increase the likelihood of being involved in a crash, especially for long glances. Recent evidence further suggests that the detrimental impact of these glances carries over and disrupting the ongoing processing of information after the eyes return to the road. This study aimed at exploring the effect of different types of visual disruptions on the top-down processes that guide the detection and monitoring of road hazards. Using a driving simulator, 56 participants were monitored with an eye tracking system while they navigated various hazardous scenarios in one of four experimental conditions: (1) Visual interruptions comprised of spatial, driving unrelated, tasks; (2) visual interruptions comprised of non-spatial, driving unrelated, tasks; (3) visual interruptions with no tasks added; and (4) no visual interruptions. In the first three conditions drivers were momentarily interrupted (either with or without a task) prior to the hazard occurrence. The visual interruption was aimed to simulate a glance inside the vehicle either with or without the need to process driving irrelevant information. Results show that the various types of tasks had differential effects on hazard detection. Implications of this study are discussed.

Drivers’ calibration in self-evaluated performance The role of task-related workload and scale specificity

Drivers tend to hold favorable or optimistic views of their skills and abilities (e.g., Horswill et al., 2004), which can lead to situations where over-confident drivers are ill-equipped (Gregersen, 1996). In addition to general self-evaluations of skills, drivers can also make erroneous estimates of their own performance and of current task demands, possibly leading to poor decisions or failures to adjust behavior to mitigate risk (e.g., Horrey et al., 2015). Gaps between perceptions, self-evaluations and objective measures have been likened to the notion of calibration.