ODU Psychologists Weigh In on Driverless Vehicles

Try conducting an internet search for "cars that drive themselves." You'll find that news media have been busy promoting the notion that in the foreseeable future we will be able to leave the driving to autonomous, computer-guided vehicles. In other words, you won't need a steering wheel, or a brake pedal.

For two Old Dominion University psychologists, this excitement is good news and not so good news.

Mark Scerbo, a human factors psychologist who studies how humans interact with advanced technologies, and Bryan Porter, an experimental psychologist who studies motorist behavior and traffic safety issues, have research backgrounds that fit well into this hot field. Media want to tap their findings for articles about autonomous vehicles.

On the other hand, the scientific opinions of Scerbo and Porter can put the brakes on the media swirl.

The international publication Automotive World Megatrends has an article in its latest edition headlined "The Mobile Office," in which it notes that "autonomous cars have long been a futuristic ideal," but that industry reports now say these self-driving vehicles can hit the market by 2020, "leaving the questions open of just how driverless a driverless car can be."

It fell to Scerbo and Porter, both of whom are quoted extensively in the article, to point out the reasons why drivers should not expect anytime soon to be able to conduct business on their computers - or, for that matter, apply nail polish - while their autonomous cars get them to work.

Porter told the magazine that even the latest technologies require a driver to monitor for situations in which automated systems struggle to cope with emergencies. "We are absolutely going to see the attempted dissemination of autonomous vehicles, throughout the U.S., Western Europe, and Australia at least, and such testing is already here," Porter acknowledged in the article. "Counterbalancing this trend will be the challenge that must be overcome before we can call the car an office."

The Megatrends article quotes a researcher at the private Transport Research Laboratory as saying the day will come soon when cars will be able to travel autonomously on certain roads under controlled conditions, allowing the driver to devote his or her attention to other things. "It will be important to ensure that if the driver is required to resume control by reaching the end of the highway or because the vehicle encounters conditions for which it is not prepared, they can do so in a manner that is timely."

This is where Scerbo expresses skepticism.

"Researchers who study ... pilot interaction with automated systems have found that they can become overly complacent in operating the aircraft," Scerbo was quoted as saying. "This can lead to several serious problems. First, they tend to accept that the automation is working properly and spend less time monitoring it. Thus, if the technology should fail they are less likely to notice it and take corrective action in a timely manner."

Scerbo said his research and that of others shows that this problem becomes more severe with more reliable systems. "Prolonged reliance on automated systems can result in skill degradation," Scerbo says in the article. "That is, when the user needs to take back control it may require a period of skill readjustments. In emergency situations, that period of readjustment may be too long to avoid an adverse outcome."

A few weeks after the Megatrends article appeared, most major media outlets in the United States and some outside the country reported on a new, two-door, electric powered car developed by Google to be autonomous. It relies on a Google Maps-related guidance system to get around.

"Google's new driverless car has no brakes or steering wheel," read the headline of a The Washington Post article.

From the headlines about the Google vehicle, one might assume that the company has found a way for vehicles to be truly driverless. But don't jump to that conclusion, said Scerbo.

He pointed out that Google has acknowledged that when it begins soon to test the cars on the open road, it will add controls so that one of Google's test drivers can take over control of the vehicle if a problem arises.

So, a human can intervene," Scerbo said in an interview. "Ironically, this is like asking the passenger to take over. The passenger is usually not engaged in driving and is poorly prepared to take over in emergency situations."

Research on intelligent vehicles and an intelligent highway system is still in its early stages, but researchers in the human factors community have known for some time that the advantages of automated systems come with consequences to the users, Scerbo said. "Some consequences are obvious but many are not. One consequence can be the loss of manual skills. An obvious example is the automatic transmission. It has made shifting gears easier for many drivers, but has also created an entire cohort of drivers who cannot operate a vehicle with a manual transmission.

"Another consequence of automation is that it often changes how tasks need to be performed and these changes can introduce hazards when operators do not fully understand the automation. Consider the 4-wheel antilock braking system (ABS) available in many automobiles today. This system was designed to increase safety by monitoring and controlling traction among the wheels thereby preventing the vehicle from locking up and skidding when the driver abruptly applies the brakes. However, the ABS changes how the driver controls the vehicle in an important way. Drivers of vehicles with traditional braking systems were taught to pump the brakes when entering a skid. However, pumping the brakes with an ABS system switches the automation on and off and undermines the functionality of the ABS. Instead, ABS requires steady pressure on the brakes. Hence, drivers who are familiar with traditional braking systems and find themselves in a vehicle with ABS, may perform an action that actually decreases their control of the vehicle under emergency conditions."

Most researchers who study automated systems agree that the user must stay engaged with operating the system at some level to minimize the potential for automation-induced errors, Scerbo said. "Unless we are talking about a system where the driver gives up control of the vehicle to another transportation system - like driving your car onto a ferryboat - the idea that drivers could put their vehicles into automatic mode and no longer need to pay attention to the road is not likely to be pursued as a viable form of automotive automation. Even the most sophisticated forms of automation in commercial cockpits today require some level of pilot engagement.