XPost: austin.general, sac.politics, alt.fan.rush-limbaugh   
   XPost: alt.google-sucks   
   From: wrd@google.com   
      
   MOUNTAIN VIEW, Calif. — Google, a leader in efforts to create   
   driverless cars, has run into an odd safety conundrum: humans.   
      
   Last month, as one of Google’s self-driving cars approached a   
   crosswalk, it did what it was supposed to do when it slowed to   
   allow a pedestrian to cross, prompting its “safety driver” to   
   apply the brakes. The pedestrian was fine, but not so much   
   Google’s car, which was hit from behind by a human-driven sedan.   
      
   Google’s fleet of autonomous test cars is programmed to follow   
   the letter of the law. But it can be tough to get around if you   
   are a stickler for the rules. One Google car, in a test in 2009,   
   couldn’t get through a four-way stop because its sensors kept   
   waiting for other (human) drivers to stop completely and let it   
   go. The human drivers kept inching forward, looking for the   
   advantage — paralyzing Google’s robot.   
      
   It is not just a Google issue. Researchers in the fledgling   
   field of autonomous vehicles say that one of the biggest   
   challenges facing automated cars is blending them into a world   
   in which humans don’t behave by the book. “The real problem is   
   that the car is too safe,” said Donald Norman, director of the   
   Design Lab at the University of California, San Diego, who   
   studies autonomous vehicles.   
      
   “They have to learn to be aggressive in the right amount, and   
   the right amount depends on the culture.”   
      
   Traffic wrecks and deaths could well plummet in a world without   
   any drivers, as some researchers predict. But wide use of self-   
   driving cars is still many years away, and testers are still   
   sorting out hypothetical risks — like hackers — and real world   
   challenges, like what happens when an autonomous car breaks down   
   on the highway.   
      
   For now, there is the nearer-term problem of blending robots and   
   humans. Already, cars from several automakers have technology   
   that can warn or even take over for a driver, whether through   
   advanced cruise control or brakes that apply themselves. Uber is   
   working on the self-driving car technology, and Google expanded   
   its tests in July to Austin, Tex.   
      
   Google cars regularly take quick, evasive maneuvers or exercise   
   caution in ways that are at once the most cautious approach, but   
   also out of step with the other vehicles on the road.   
      
   “It’s always going to follow the rules, I mean, almost to a   
   point where human drivers who get in the car and are like ‘Why   
   is the car doing that?’” said Tom Supple, a Google safety driver   
   during a recent test drive on the streets near Google’s Silicon   
   Valley headquarters.   
      
   Since 2009, Google cars have been in 16 crashes, mostly fender-   
   benders, and in every single case, the company says, a human was   
   at fault. This includes the rear-ender crash on Aug. 20, and   
   reported Tuesday by Google. The Google car slowed for a   
   pedestrian, then the Google employee manually applied the   
   brakes. The car was hit from behind, sending the employee to the   
   emergency room for mild whiplash.   
      
   Google’s report on the incident adds another twist: While the   
   safety driver did the right thing by applying the brakes, if the   
   autonomous car had been left alone, it might have braked less   
   hard and traveled closer to the crosswalk, giving the car behind   
   a little more room to stop. Would that have prevented the   
   collision? Google says it’s impossible to say.   
      
   There was a single case in which Google says the company was   
   responsible for a crash. It happened in August 2011, when one of   
   its Google cars collided with another moving vehicle. But,   
   remarkably, the Google car was being piloted at the time by an   
   employee. Another human at fault.   
      
   Humans and machines, it seems, are an imperfect mix. Take lane   
   departure technology, which uses a beep or steering-wheel   
   vibration to warn a driver if the car drifts into another lane.   
   A 2012 insurance industry study that surprised researchers found   
   that cars with these systems experienced a slightly higher crash   
   rate than cars without them.   
      
   Bill Windsor, a safety expert with Nationwide Insurance, said   
   that drivers who grew irritated by the beep might turn the   
   system off. That highlights a clash between the way humans   
   actually behave and how the cars wrongly interpret that   
   behavior; the car beeps when a driver moves into another lane   
   but, in reality, the human driver is intending to change lanes   
   without having signaled so the driver, irked by the beep, turns   
   the technology off.   
      
   Mr. Windsor recently experienced firsthand one of the challenges   
   as sophisticated car technology clashes with actual human   
   behavior. He was on a road trip in his new Volvo, which comes   
   equipped with “adaptive cruise control.” The technology causes   
   the car to automatically adapt its speeds when traffic   
   conditions warrant.   
      
   But the technology, like Google’s car, drives by the book. It   
   leaves what is considered the safe distance between itself and   
   the car ahead. This also happens to be enough space for a car in   
   an adjoining lane to squeeze into, and, Mr. Windsor said, they   
   often tried.   
      
   Dmitri Dolgov, head of software for Google’s Self-Driving Car   
   Project, said that one thing he had learned from the project was   
   that human drivers needed to be “less idiotic.”   
      
   On a recent outing with New York Times journalists, the Google   
   driverless car took two evasive maneuvers that simultaneously   
   displayed how the car errs on the cautious side, but also how   
   jarring that experience can be. In one maneuver, it swerved   
   sharply in a residential neighborhood to avoid a car that was   
   poorly parked, so much so that the Google sensors couldn’t tell   
   if it might pull into traffic.   
      
   More jarring for human passengers was a maneuver that the Google   
   car took as it approached a red light in moderate traffic. The   
   laser system mounted on top of the driverless car sensed that a   
   vehicle coming the other direction was approaching the red light   
   at higher-than-safe speeds. The Google car immediately jerked to   
   the right in case it had to avoid a collision. In the end, the   
   oncoming car was just doing what human drivers so often do: not   
   approach a red light cautiously enough, though the driver did   
   stop well in time.   
      
   Courtney Hohne, a spokeswoman for the Google project, said   
   current testing was devoted to “smoothing out” the relationship   
   between the car’s software and humans. For instance, at four-way   
   stops, the program lets the car inch forward, as the rest of us   
   might, asserting its turn while looking for signs that it is   
   being allowed to go.   
      
   The way humans often deal with these situations is that “they   
   make eye contact. On the fly, they make agreements about who has   
   the right of way,” said John Lee, a professor of industrial and   
   systems engineering and expert in driver safety and automation   
   at the University of Wisconsin.   
      
   “Where are the eyes in an autonomous vehicle?” he added.   
      
   But Mr. Norman, from the design center in San Diego, after years   
   of urging caution on driverless cars, now welcomes quick   
   adoption because he says other motorists are increasingly   
      
   [continued in next message]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)