قالب وردپرس درنا توس
Home / Business / Self-propelled Uber had no emergency braking turned on when it reached pedestrians, says NTSB

Self-propelled Uber had no emergency braking turned on when it reached pedestrians, says NTSB




An image of a mounted video camera nearing a self-driving Uber SUV hit a woman in Tempe, Arizona, in March. (Tempe Police Department / AP)

The self-driving Uber, which attacked and killed a pedestrian in March, first identified the woman as a vehicle and deliberately put her on the road without the emergency brake system on.

"According to Uber, emergency braking maneuvers are not possible while the vehicle is under computer control to reduce the potential for unpredictable vehicle behavior," according to a preliminary report from the National Transportation Safety Board

Human safety The driver only began when the 49-year-old Elaine Herzberg was hit while pushing a bicycle across a dark road from Tempe, Ariz., the NTSB said.

Federal safety investigators have not given a reason for the crash. But they described initial findings that raised far-reaching questions about Uber's decision-making, engineering and safety concept while developing a potentially lucrative driverless system on public roads.

Uber's Volvo XC90, which killed Herzberg, comes equipped with an automatic emergency braking function called City Safety, according to NTSB. Uber disables this and some other security features when the car is controlled by Uber's self-driving system, but stops when the car is manually driven by one person.

Uber has stated that the Volvo crash reduction systems are designed to support drivers, not part of a self-driving system

On questions, a Uber spokeswoman makes a statement in which no concrete results are addressed. It said the company has worked closely with the NTSB and is reviewing the security of its program. "We also got former NTSB chairman Christopher Hart to talk to us about our overall safety culture, and we look forward to reporting more about the changes in the coming weeks," the statement said.

The NTSB, the Uber SUV's sensors, first recognized Herzberg "about 6 seconds before the impact, when the vehicle was traveling at 43 mph." As the vehicle and pedestrian routes converged, the self-propelled system software classified the pedestrian as an unknown object, as a vehicle, and then as a bicycle with different expectations of the future driveway. "

Then beat 1.3 seconds before Herzberg," the self-propelled system found that an emergency braking maneuver was needed to mitigate a collision, "the NTSB said, but since Uber's emergency braking maneuvers were" not allowed. " The NTSB, without comment, outlines the inherent separation in Uber's procedure.

"The vehicle operator is instructed to intervene and take action. The system is not designed to warn the driver, "says the preliminary report.

One reason why Uber would have disabled automatic emergency braking in its self-driving cars is to avoid cameras or sensors that may pose potential problems Brakes, experts said UBS wanted to reduce Uber's "unpredictable vehicle behavior."

Uber may have tried to "reduce the number of false positives when the computer misclassifies a situation and automatic emergency braking intervenes unnecessarily," said About Costa Samaras, robotics expert and assistant engineer at Carnegie Mellon University in Pittsburgh. "False positives like these could also be dangerous, especially at higher speeds."

Samaras said that in this case it's false-positive Results seemed to give a false negative.

"The car has seized the pedestrian six seconds earlier but they misclassified them up to 1.3 seconds before the impact, "he said. "Even then, the computer found that emergency braking was required but the function was disabled and there is no mechanism to warn the driver."

"We know that humans are a terrible security system, we are easily distracted and we have slower reaction times," Samaras added. "Making the driver aware of such situations before the crash seems like child's play."

Rafaela Vasquez, the security driver, is seen in a Tempe police video that looked down several times just before the crash. She said she looked at elements of Uber's self-propelled system and not a cell phone she said she did not need until she called 911. The NTSB further investigates these and other elements of the crash, adding that "the operator is responsible for monitoring diagnostic messages appearing on an interface in the center stack of the vehicle and highlighting events of interest for subsequent review."

Vasquez was not tested for alcohol or drugs, but the police said she showed no signs of impairment. 19659019] The victim, Herzberg, was tested positive for marijuana and methamphetamine according to NTSB. She wore dark clothes and walked in front of the crosswalk. The bike she pushed had reflectors, the NTSB said, but they pointed at the oncoming Uber.

Data found after the accident "showed that all aspects of the self-propelled system were working normally at the time of the crash," investigators said, and there were no error messages.

Investigators said Uber's system is not designed to warn safety drivers that they should stop the car in such situations. This decision reflects shortcomings in other areas in recent decades, as people increasingly rely on automation, according to Duke University robotics expert Missy Cummings.

"This lesson has been written with blood over and over again," said Cummings, director of the University's Humans and Autonomy Lab. She cited Three Mile Island's nuclear accident in 1979 and several plane crashes and said that under these circumstances, engineers decided not to give the human operators critical information they would need to prevent tragedies.

Cummings said computer vision also remains problematic and often has difficulty figuring out which items on the road represent actual dangers, sometimes through something so innocent like a plastic bag confused. And the communication between vision and braking systems can be fuzzy and fail for a variety of reasons, she said. "That's why we have to be careful about putting these cars on the road before we work out these issues," Cummings said.

Uber announced this week that it will terminate its self-propelled operation in Arizona and almost come to rest 300 workers, mostly backup drivers. The company has spoken with state and local officials in Pennsylvania, hoping to resume public road tests this summer in Pittsburgh, where the driverless research group is based.

But Uber has angered Pittsburgh's mayor William Peduto (D), an enthusiastic early Uber supporter who now emphasizes safety concerns. He has, for example, demanded that the company agree to limit their self-driving cars to 25 miles per hour in the city, regardless of the specified speed limit, as slower speeds increase the chance that pedestrians will survive.

He responded to the conditions he had raised in the talks, "said Timothy McNulty, Peduto's spokesman.

Driverless Car Law, passed by the US House of Representatives last year, did not allow the communities In Washington, automakers and driverless tech companies have cracked down on what they could warn a crippling "patchwork" of state and local regulations that would limit the economic and safety benefits of driverless automobiles.

If Uber Local Approves Demands Voluntarily the restrictions would not be dependent on state or federal statutes, "McNulty said.


Source link