The benefits and pitfalls of advanced driver assistance systems for mitigating drowsy driving crashes

As advanced driver assistance systems that assist in managing the risk of drowsiness-related motor vehicle crashes rapidly evolve, sleep clinicians, scientists, and regulators need to understand their role and limitations, and contribute to their evaluation and regulation. Drowsiness continues to be a key cause of road crashes that result in more than 1.2 million fatalities per year and 50 times as many injuries internationally [1]. It is important to determine which vehicle technologies can mitigate different types of crashes, in different driver populations and vehicles. Heavy vehicles are overrepresented in fatal collisions, with their drivers at particular risk of drowsiness-related crashes due to shiftwork conditions, high risk of sleep disorders, and driving exposure. Using a large real-world dataset, Kawaguchi et al demonstrated that while autonomous emergency braking systems did not reduce drowsiness-related crashes in heavy vehicles they did reduce non-drowsiness-related crashes [2]. Real-world data suggest AEB systems can reduce total road crashes by between 6% and 18% and injuries and fatalities by 7% [3]. Rear-end collisions in cars appear to be reduced by 25%–50% [3]. However, the effectiveness of AEBs is influenced by a range of factors. They are designed and tested for effectiveness at lower speeds of up to 80 kph [4]. Weather and road conditions may alter their ability to detect oncoming hazards and/or reduce the effectiveness of braking generally. As pointed out by Kawaguchi, trucks have a larger mass requiring longer braking distances, and frequently operate at high speeds, reducing the time available for oncoming hazard detection. When a driver is impacted by drowsiness their response to hazards and vehicle technology warnings is delayed or absent, with failure to correct steering or brake so that crashes occur at higher speeds. Kawaguchi et al rightly point out that, while AEB technology is designed to mitigate rear-end crashes, it is less effective in other types of crash configurations. In the United States, the overwhelming majority of drowsy-related crashes involve head-on collisions and road departures [5]. The current study suggests that this combination of factors impairs the effectiveness of AEBs for preventing drowsiness-related crashes, at least with current technology. Of note, the cost of drowsiness-related collisions was double that of other collisions, as expected with higher speed collisions, and AEBs did not mitigate the cost of drowsiness-related collisions [2]. There are limitations in the study being retrospective and the collision rate based on collisions per truck per year without taking driving time or distance into account. Nevertheless, this large real-world data include an objective assessment of drowsiness and cost and is a valuable contribution to understanding the limitations of AEBs in drowsiness-related truck collisions. The objective drowsiness monitoring utilized driver-facing video to assess periods of eye closure. Prolonged eye closure is a good indicator of driver drowsiness, although standardized metrics such as PERCLOS (percent of time with eyes closed) were not used, which might have influenced the accuracy of drowsiness assessment [6]. Strategies to reduce the likelihood of truck drivers driving while drowsy such as optimizing shift schedule design and regulation and sleep disorders screening, of course, remain a key component of minimizing risk. However, it is not possible to eliminate the presence of drowsy driving completely through these strategies, as undiagnosed sleep disorders, driving at night during the body’s circadian nadir, and periods of acute and chronic sleep restriction will remain in this shift work population. Drivers often continue to drive, even when presented with break opportunities and they are aware of their level of drowsiness [7]. This underpins the importance of continuing to reduce the risk of drowsiness-related collisions occurring, and mitigating the likelihood of injuries and fatalities through improving vehicle technology and road infrastructure as well. Driver monitoring systems (DMS) that have good accuracy for detecting drowsiness and distraction through measurement of eyelid closure and direction of gaze are becoming widespread in the transport sector and have been mandated in new passenger vehicles in Europe [6, 8]. Limited field studies have found that in cabin DMS drowsiness alerts substantially reduce the occurrence of drowsiness events and improve driving performance [9, 10]. Other advanced driver assistance systems, such as lane departure warnings, lane keeping assist, and automated driving systems that take over vehicle operation are also becoming widespread. Given drowsiness-related crashes often result from lane and road departures, these technologies are likely to reduce drowsiness-related crashes and injuries, but their effectiveness and limitations in different populations need to be further defined. Initially intended as a last line of defense to alert drivers or take over vehicle control when a critical event is detected, the technology should be combined with effective strategies to mitigate drowsiness such as breaks and caffeine [11]. When driver alerts are combined with alert notifications to driver supervisors in occupational settings this enhances the reduction in drowsiness events, perhaps through promoting these behaviors [10]. It is unclear whether they promote other preventive drowsiness driver behaviors, such as increased pre-drive sleep and strategic napping. Frequent alerts, especially those considered to be false, often become ignored by operators or lead to slowed responses [12, 13]. It follows that alerts need to be accurate and reliable. Drivers with sleep disorders are being identified through vehicle technologies such as DMS as well as through other sleep disorders screening methods. Kawaguchi et al found a higher body mass index, a key sleep apnea risk factor, in the drivers who had had drowsiness-related collisions, where drowsiness was defined by video-based assessment of the drivers [2]. There are currently no guidelines regarding when drivers with frequent alerts from DMS for drowsiness, lane departure or AEB activation should be assessed for sleep disorders. DMS can identify people with sleep disorders and regular drowsiness while driving [14]. They could also play a role in monitoring the effectiveness of treatment for sleep disorders regards driving safety. As these technologies become widespread it is critical for sleep scientists, clinicians, technology companies and regulators to collaborate and develop criteria for sleep disorders referral when drivers receive regular driving performance or drowsiness alerts. Analysis of large data sets from driver assistance and automated driving systems can also be used to identify high-risk driving schedules and routes [15]. This has the potential to inform schedule changes for individual companies and driving hours regulations. Sleep clinicians will be increasingly referred drivers with driving impairment and/or drowsiness identified through vehicle technology and asked to provide advice regards their fitness to drive based on the technology alerts. There is great potential, but much to learn about the role, benefits, and pitfalls of these technologies. It is time for sleep scientists and clinicians to engage with vehicle technology manufacturers and regulators to chart the road ahead! Financial disclosure: No funding was received for this work. MEH is employed through the Institute for Breathing and Sleep and Austin Health. WJH is employed through the AAA Foundation for Traffic Safety. MEH has previously received research grant support from the Co-operative Research Centre for Alertness, Safety and Productivity. Nonfinancial disclosure: MEH has previously received equipment support for research projects from Optalert and Seeing Machines No new data were generated or analysed in support of this research.