First appeared in NY Times
A DRIVERLESS car is not yet ready for the market. But in the meantime, automakers are continuing to market some components of one. The Ford Motor Company announced last month that it would offer “lane-keeping technology” as an option for its 2013 Ford Fusion and Ford Explorer. The price has not yet been set.
When lane-keeping technology works, it can save lives. But it is suited only for certain road conditions, and there are reasons to doubt that it will activate as consistently as it should.
Ford’s technology relies on a camera mounted to the rear-view mirror. When the system is switched on and the vehicle is traveling more than 40 miles per hour, it will use the road’s lane markings to sense veering near one edge of the lane or the other. If the turn signal is off, the system will assume that the drift is unintentional and will send a vibration to the steering wheel as a warning.
If the driver doesn’t correct the drift, the software is then supposed to engage the power steering and turn the car back toward the center of the lane.
When all goes well, this will be flat-out wonderful. But the camera may have difficulty detecting the lane markings — when the sun is at a low angle, for example, or during heavy rainfall or on curves. If it fails to see the markings, it simply remains dormant.
The National Highway Traffic Safety Administration has declined to give lane-keeping technology its blessing. Ronald Medford, deputy administrator of the agency, says, “We believe additional evaluation and research about lane-keeping systems is needed before we can decide whether we should recommend it to the public.”
Earlier research by the highway agency found performance problems in other systems then available. When asked about that, Michael Kane, a development engineer at Ford, acknowledges that certain conditions, such as when driving into direct sunlight, confound the system. “It’s not perfect,” he says.
But, he continues, “we’ve worked to enable the system to detect lane markings on a much higher percentage of situations, such as tree-lined curves with lots of shadows.”
When Toyota introduced similar technology in its 2010 Prius, it chose to call it “Lane Keep Assist.” Lexus and Mercedes also use similar language for their systems. That “assist” keeps expectations from getting out of hand. Ford is going instead with “Lane Keeping System,” without any namby-pamby qualifier.
Advanced automotive safety technologies, like lane-keeping systems, are most prevalent in Europe, says Eddy Llaneras, a staff member of the Virginia Tech Transportation Institute.
“In Europe, the driving population understands that ultimate responsibility remains with the driver,” he says. “Here in the U.S., we are inclined to believe the technology will do all the work for us. And if it doesn’t, we’ll sue.”
Mr. Llaneras expects that within five to 10 years we will see “lane centering,” in which the vehicle steers itself. “The technology exists today — it’s been tested on research vehicles,” he says. “But manufacturers hesitate to introduce it because the systems can never be 100 percent reliable and they introduce liability exposure.”
Mercedes makes its lane-keeping system active by default, but Ford, like Toyota and Lexus, insists that its system be turned on manually, every time; the driver doesn’t have the option to designate that it be activated by default, or when cruise control is used — a seemingly natural combination.
Ford’s new technology package also includes what it calls a “Driver Alert System,” which will provide warnings when the software detects a pattern of driving associated with drowsiness, such as weaving within the lane boundaries. For the first warning, Ford’s designers have chosen to sound a soft chime and to display this message on the dashboard: “Rest suggested.” If the problem persists, a louder chime and a new message — “Rest now” — follow.
Mr. Kane could not point to any test data, however, that suggests that a driver who is sleepy to the point of being dangerous is likely to find and heed display warnings. Notwithstanding the chimes, delivering the message with an insistent recorded voice would seem a better choice.
J. Christian Gerdes, director of the Center for Automotive Research at Stanford and an associate professor of mechanical engineering, has been testing driverless cars since the early 1990s. One such vehicle successfully climbed the twisty road to the top of Pikes Peak in Colorado — and did so multiple times. Its technology, however, isn’t coming soon to an auto showroom near you: its navigation relies on a base station in the area to transmit a Differential Global Positioning System signal.
Even if camera-based lane-keeping systems were to work well, we might not end up safer. Mr. Gerdes calls attention to the “risk accommodation” problem. “As vehicles are made safer, drivers may compensate by engaging in riskier behavior,” he explains.
As humans, we have one thing that works in our favor while driving: we are more likely to handle unexpected events successfully than the software in an autonomous car.
“Humans understand context,” Mr. Gerdes says. “If I’m driving along a row of parked cars and a ball rolls out into the street, I know to look for a child.”
Then again, we humans aren’t always exemplary in paying attention to the road. Ford’s system may lack perfect vision, but it will never be distracted because it’s checking its e-mail.