An ordinary car has about 30,000 separate parts, but only one component is persistently prone to catastrophic failure: the driver. Whereas 2 percent of accidents are caused by equipment malfunction, 94 percent are the driver’s fault.
That is why much of the progress in highway safety during the past century has resulted from behavioral science that reveals how drivers interact with their vehicles. The value of this work will only increase as the nation finds itself on the verge of a revolution in personal transportation—the self-driving automobile.
Many safety systems and procedures we take for granted today arose from applied behavioral research—where cognitive science meets engineering, and both advance in tandem. For example, the fact that cars are now equipped with a center-mounted supplementary brake light more easily seen by drivers following behind is thanks to a groundbreaking study of rear-end crashes conducted by California psychologist John Voevodsky in the early 1970s.
Voevodsky was a pioneer in what is now known as the science of human factors and ergonomics. The field had been growing rapidly since World War II, when there was a sudden pressing need to reduce errors and accidents caused by personnel using complicated military systems. Understanding how operators act and designing methods to optimize safety and system performance became increasingly urgent priorities. After the war, as the number of vehicles and drivers rose rapidly, automobile operation became a natural focus for the field.
Human factors, indeed any science, works like a jigsaw puzzle: Each piece of new knowledge helps reveal the shape of the overall pattern and then shows where the next piece needs to go. When Voevodsky began his investigations, many pieces were already in place. In 1955, the Society of Automotive Engineers had collected and published anthropometric data for people ranging in size from the 5th to 95th percentile. Standardized measurement procedures arose and by the early 1960s behavioral science had produced a new tool called the “eyellipse.” It was a set of specifications, based on the location of a driver’s eyes within a car, exactly quantifying the ellipse-shaped field of view available to people of different shapes and heights—including the neck pivot points at which vision is cut off. Those advances provided a basis for evaluating what drivers could and could not see.
Voevodsky went beyond this focus, concentrating on what the drivers neglected to see. Observing the frequency of life-threatening rear-end crashes, which then made up half of all traffic accidents (closer to one quarter in recent years), he hypothesized that an extra warning light, mounted between the two traditional brake lights and designed to flash faster as the car slowed down, would better capture a following driver’s attention, reducing the risk of collision.
He equipped 343 San Francisco taxicabs with these extra lights; a control group of 160 cabs had none. After 12.3 million miles of normal driving, rear-end collisions in the test group were reduced by 60.6% compared to the control group. In 1974 in the Journal of Applied Psychology Voevodsky reported that the warning light had prevented 5.4 collisions, 1.2 cab driver injuries, and $643 ($3,262 in 2016 dollars) of taxicab damage per million miles.
The National Highway Traffic Safety Administration (NHTSA) subsequently conducted larger follow-up studies, leading to its 1986 decision to require that all new automobiles be manufactured with a high-positioned third brake light.
Nonetheless, 30 years later rear-end collisions caused by driver inattention remain a severe national problem, exacerbated by the proliferation of in-car distractions from cell phones, dashboard display screens, on-board navigation systems, email and social media connections, and more.
A recent large-scale study found that “potentially 36%, or 4 million, of the nearly 11 million crashes occurring in the United States annually could be avoided if no distraction was present.” Compared to an attentive, undistracted driver, the data shows that operating the car’s radio roughly doubles the risk of a crash, while using touch-screen menus increase it by a factor of 4.6. Texting makes an accident 6.1 times more likely, reaching for an object or reading/writing raises the risk by a factor of 9, and dialing a cell phone by a factor of 12.2—the highest of any distraction observed.
Technologies have arisen to mitigate these risks, but human factors research has highlighted shortcomings. A case in point: voice-operated, “hands-free” communication systems. As early as 2001, behavioral engineer John D. Lee, now at the University of Wisconsin, and colleagues found that the use of speech-based email systems increased driver reaction time by about 30 percent. In 2014, the National Safety Council reported, use of any kind of cell phone, for texting or speech, hand-held or “hands-free,” was involved in 26 percent of all motor vehicle crashes.
“Talking to your car is horribly demanding,” says David L. Strayer, a professor of psychology at the University of Utah who studies the mental effort expended by drivers in various situations. “It far eclipses the mental workload of talking to a person right next to you.” Compounding the problem are the various ways in which auto manufacturers implement voice-activated systems. Strayer’s group created a five-point driver-distraction scale for the American Automobile Association’s (AAA’s) Foundation for Traffic Safety, and found that performing exactly the same task ranged from a distraction rating of 2.4 to 4.6 depending on the car’s make and model.
In the face of chronic operator inattention, researchers have developed sensor-activated automated systems, such as automatic emergency braking (AEB), front-facing radar, blind-spot monitors, and lane-departure warning systems, to compensate for driver distraction. Human factors research has played a major role in the design of these systems.
Repeated studies indicate that average driver response time (often defined as the period between perceiving a danger and applying the brakes) ranges from less than 0.5 seconds to more than 2.5 seconds depending on the driver and the situation. AEB systems, rapidly becoming standard equipment on many vehicles, have to be deployed within those constraints.
But once a threat is detected by sensors, how soon should warnings be delivered? “Too early and drivers receive annoying false alarms,” Lee says, “too late and they might not be able to avoid crashing.” Using a high-fidelity driving simulator, Lee and colleagues found that a rear-end collision avoidance system with an early warning in the range of 2.8 seconds reduced the number of collisions by 80.7 percent.
Human factors science also examined a related question: By what means should the driver be warned? Researchers compared the performance of visual, auditory, and tactile warnings when the driver was talking on a cell phone. A loud, high-pitched tone was effective, but a sudden vibration in the steering wheel or in a vibrator attached to the driver’s seat belt prompted driver reaction earlier—perhaps because talking on the phone diminished the driver’s auditory attention.
Today, driver inattention—and ways to reduce it or compensate for it— have become one of the most studied topics in human factors, and behavioral science is producing a prodigious volume of data on driver reaction to AEB and other features.
One impact of recent research was evident in March 2016 when the NHSTA and the Insurance Institute for Highway Safety (IIHS) announced an “historic commitment by 20 automakers representing more than 99 percent of the U.S. auto market to make automatic emergency braking (AEB) a standard feature on virtually all new cars”—no later than 2022. The IIHS estimates that the move “will make AEB standard on new cars 3 years faster than could be achieved through the formal regulatory process. During those 3 years, according to IIHS estimates, the commitment will prevent 28,000 crashes and 12,000 injuries.”
That automakers’ agreement in no way diminishes the need for continuing human factors research. The long-term success of AEB and other coming innovations will depend critically on how drivers react to them. One issue will be minimizing the effect called “automation surprise” in which an automated system performs an operation totally unexpected by the user, with potentially disastrous results.
Indeed, human factors science will soon be facing one of the most daunting challenges in its history with the fast-approaching advent of self-driving cars, or autonomous vehicles (AVs). In September 2016, the U.S. Department of Transportation issued its Federal Automated Vehicles Policy which “sets out an ambitious approach to accelerate the [AV] revolution,” including “performance guidance” for system developers.
Rigorous testing will be crucial. But it will have to include more than on-road performance data. The RAND Corporation notes in a new report that “autonomous vehicles would have to be driven hundreds of millions of miles and sometimes hundreds of billions of miles to demonstrate their reliability,” a process that could take tens or even hundreds of years to complete. “Therefore, at least for fatalities and injuries, test-driving alone cannot provide sufficient evidence for demonstrating autonomous vehicle safety.”
Less than exhaustive reliability testing will place new demands on human factors scientists. A typical AV might rely on a hugely complex integrated system of on-board cameras, lidar, GPS signals, and maps stored in the vehicle’s memory—and perhaps eventually also on vehicle-to-vehicle electronic communication. Optimal design—especially during the transition period in which cars are only partially self-driving—will demand investigation of the range of possible driver reactions in different degrees of human control. Scientists have only begun to explore this formidable range of complications.
“When automation is perceived as proficient,” two engineers write in the newsletter of the National Academies of Sciences, Engineering, and Medicine’s Transportation Research Board, “operators rely more heavily on the technology and fail to use their own skills. This leads to a loss of skill and increases reliance on the automation, possibly leading back to mode confusion,” which occurs when the operator may “make decisions believing that the system is in a different state than it actually is.”
Lee concurs: “Systems must be designed so that drivers develop appropriate levels of trust and accurate expectations of the automation. If the driver thinks the car can drive itself, the lure of distractions will quickly pull their eyes off the road and hands off the wheel.”
Finally, behavioral science may also have to be applied “in reverse” to ensure that fully automated control systems act more like real people. Few human drivers follow all the rules, and AVs programmed to abide strictly by such rules can be paralyzed by routine human behaviors such as jockeying for position at a four-way stop or swerving into the space behind a car even when it is closer than the canonical one car length for every 10 mph of speed. Some such situations can trigger immediate braking by an AV, raising the risk that it will be struck from behind.
For the current generation of AVs, it is possible that “the real problem is that the car is too safe,” Donald Norman, director of The Design Lab at the University of California, San Diego, who studies AVs, told The New York Times. “They have to learn to be aggressive in the right amount, and the right amount depends on the culture.”
This article was written by Curt Suplee for From Research to Reward, a series produced by the National Academy of Sciences. This and other articles in the series can be found at www.nasonline.org/r2r. The Academy, located in Washington, DC, is a society of distinguished scholars dedicated to the use of science and technology for the public welfare. For more than 150 years, it has provided independent, objective scientific advice to the nation.
Please direct comments or questions about this series to Stephen Mautner at email@example.com.
Photo and Illustration Credits:
Traffic on a rainy night: Dubravko Grakalic/Alamy Stock Photo • Operating an onboard nav system: MARKA/Alamy Stock Photo • Handheld smartphone: Lev Dolgachov/Alamy Stock Photo • National Advanced Driving Simulator: Bill Nellans • John D. Lee with students: University of Wisconsin-Madison • David Strayer: Andy Wiebe
© 2016 by the National Academy of Sciences. These materials may be reposted and reproduced solely for non-commercial educational use without the written permission of the National Academy of Sciences.