On the road but not alert: Is tech to blame?
SAN FRANCISCO – On a busy stretch of highway somewhere, a driver is gazing out the window, hands on her lap, feet off the pedals. Such driver-assisted motoring, a midpoint on the journey to fully self-driving cars, uses radar and cameras to help a car steer, brake and even change lanes. As such features emerge in less-expensive cars, a vexing question looms: Automakers from Tesla to Nissan caution that their tech must be monitored, but can humans be trusted to do so?
“These systems are designed not only for ideal environments like good weather and clear lane markings but also for rational behavior, and humans are predictably irrational,” says Bryan Reimer, a research scientist at the Massachusetts Institute of Technology who specializes in self-driving tech. “The question is, does the automation work so well and require so little of you that in essence, a new safety problem is created while solving for another?”
Two crashes spotlighted such concerns. A month ago, an Uber self-driving car killed a pedestrian in Arizona despite having a driver at the wheel who was supposed to monitor the vehicle. The incident ignited debate over whether testing autonomous cars in public is appropriate.
A week later, a Tesla Model X in autopilot mode slammed into a concrete highway divider in Mountain View, Calif., killing driver Walter Huang. Huang’s family sued Tesla, but the company says Huang repeatedly ignored prompts to retake control of the wheel.
As litigators fight over blame some experts wonder whether such features encourage drivers to take more risks than the technology can handle.
Nearly a dozen automakers offer variations of this tech, from MercedesBenz’s Driver Assistance Package on its $90,000-and-up S Class sedan to Nissan’s ProPilot Assist available on the $23,000-and-up Altima.
Dubbed Level 2 systems by the Society of Automotive Engineers (Level 1 being cruise control and Level 5 being autonomous mode), these technology packages typically require drivers to check in with the system through frequent steering wheel inputs.
“In every system I’ve tried, I’ve been able to discern situations where (the car) got confused, such as when lane lines drift away at an on-ramp,” says Karl Brauer, executive publisher of Cox Automotive. “If you have this tech, realize there are common circumstances even the best system can’t handle. If you can’t watch out for that, you shouldn’t have it.”
One of the most innovative and ambitious driver-assist systems on the market comes from General Motors, whose Super Cruise package lets drivers take hands off the wheel and feet off the pedals as the vehicle steers itself along any highway GM has mapped into the car’s computer brain.
Super Cruise creates a virtual center line in the lane that the car follows, and its mapped knowledge of the road’s bends and elevation changes keeps the vehicle at a steady speed.
Using a driver-facing infrared camera that monitors head position, the system will hassle a driver who is not looking straight ahead.
“(We) require the driver to be ready to take control of the vehicle at all times during operation,” says Daryl Wilson, lead development engineer on Super Cruise.
Despite the litany of warnings, humans don’t always listen.
Online videos show drivers showing off what they can do while their car motors along on its own. Some hop into the rear seats, ignoring visual and audio prompts to resume control. Some even pay for help tuning out such warnings. For $179, a plug-in device called Autopilot Buddy promises to “extend the use of Autosteer on Tesla vehicles by reducing the check-in ‘nag’ warnings.”
For every vehicle that saves the day by braking before the driver sees the threat, there are instances in which drivers get too comfortable that their vehicles have the situation in hand.
An Esurance poll found that 25% of buyers of driver-assist tech disabled at least one feature since buying the car, and 29% said they found the in-car warnings a distraction.