Autonomous vehicles promise a future of almost unimaginable driver convenience. After all, who wouldn't want to be able to watch a movie during a tedious stop-and-go rush hour commute? But proponents of AVs think the technology’s even more meaningful impact will be in the realm of safety, given that more than 90% of traffic accidents are thought to be the result of human error. Indeed, we’re already seeing some of the safety benefits as semi-autonomous driver assist features, like lane keep assist, blind spot monitoring and emergency braking, become commonplace on newer vehicles. Even so, the next big developmental leap to so-called Level 3 vehicles, which are capable of doing all the driving for us in certain situations and are now starting to enter American and European markets, could introduce a whole new set of safety and driver behavior challenges, says Associate Professor of Industrial and Manufacturing Systems Engineering Bochen Jia. That’s because Level 3 AVs still require full driver intervention when conditions are beyond the system’s capability — like when lane markers aren’t clearly marked or are obscured by weather; or in the event of system malfunction. That introduces a whole new type of driver experience, which the AV research community calls “takeover.” That’s the tricky liminal moment when a driver has to abruptly shift their focus from, say, watching their favorite Netflix series, to dealing with a potential road emergency.
Intuitively, takeover is a fraught, psychologically complex space, which is why Jia says a lot of research has been devoted to studying how drivers respond to this type of demand. For example, early research has indicated, somewhat counterintuitively, that the risk of driver fatigue actually increases when operating a Level 3 vehicle. It seems the mental task of constantly monitoring a vehicle doing the driving for us can actually be more draining than doing the driving ourselves. And in Jia’s lab, he’s been focusing on a neglected area of takeover research: younger, inexperienced drivers. From a research perspective, he says it’s a fascinating population. First, young people’s brains are still developing, particularly when it comes to attention resources and situational awareness, which can influence things like fatigue development and how well they respond to potentially hazardous situations, like takeover. In addition, younger people tend to, in general, be more trusting of technology, which could lead to less intense monitoring of the vehicle and undesirable takeover outcomes. Moreover, Jia says we might not be too far away from an era where young drivers’ only driving experience is on Level 3 vehicles. “That would mean they actually aren’t getting much experience doing actual driving, which could influence how well they respond to situations when the vehicle asks them to take over,” Jia says.
For one of his latest projects, Jia has been digging into the particular issue of fatigue onset in younger, novice drivers. For this study, he recruited both drivers under the age of 19 who had limited driving experience and middle-aged drivers who had years of experience. Using a high-fidelity driving simulator, both sets of participants engaged in extended periods of autonomous driving and were instructed to take control of the vehicle when the autonomous mode was not safe or had failed. During their driving experience, Jia then introduced several types of takeover events to see how drivers would react. By combining physiological measures and self-reported assessments collected periodically during their driving task, Jia then monitored each set of participants for fatigue onset. The results were pretty interesting: Novice young adult drivers developed both physical and mental fatigue earlier than their middle-aged counterparts during extended periods of autonomous driving. "This is somewhat surprising to people because they think young people naturally have more energy," Jia explains. "However, it appears that the mental demands of autonomous driving, or the tendency of younger drivers to be more engaged in non-driving tasks during autonomous operation, may accelerate the onset of fatigue in younger drivers." Moreover, the takeover outcomes for fatigued drivers were often far from desirable. “Essentially, we were looking at how often participants would fail to prevent a collision,” Jia says. “And it happened more frequently than we had hoped.”
Jia says the results point to a need for highly effective fatigue warning systems for Level 3 AVs that have special considerations for younger, inexperienced drivers. Interestingly, Jia envisions this system would be most effective if it wasn’t just part of the vehicle, but integrated with other readily available wellness monitoring technologies, like wearable fitness trackers that monitor heart rates and sleep quality. “I don't think you can have a system that’s confined to the car — that’s not enough,” Jia says. “It should be part of your life monitoring — an overall system — because they don’t suddenly become a different person when they get in the car. With all the personal data monitoring technologies we have nowadays, we could easily identify younger drivers, and we likely have a lot of the data we would need.”
Next up, Jia says he plans to continue his research on fatigue onset with a study that puts participants in real automated vehicles within a closed testbed, such as the Mcity test facility. In addition, he has a similar project that is studying differences in distraction levels between younger, novice drivers and middle-aged experienced drivers. He hopes both studies will provide critical insights into improving the safety and usability of autonomous vehicle systems for drivers of all experience levels.
###
Story by Lou Blouin. Want to learn more about Jia's work on driver fatigue? You can read the full-text of his recent article on the subject in the “Journal of Safety Research,” which was co-authored by Jia's student Yourui Tong, UM-Dearborn Associate Professor of Industrial and Manufacturing Systems Engineering Shan Bao, Tsinghua University Professor of Industrial Engineering Changxu Wu and UM-Dearborn Associate Professor of Psychology Nitya Sethuraman.
