• UberPeople.NET - Independent community of rideshare drivers. It's FREE to be a person and enjoy all the benefits of membership. JOIN US! CLICK HERE

Competing With the Giants in Race to Build Self-Driving Cars - NY Times 4 Jan

Michael - Cleveland

Well-Known Member
Competing With the Giants in Race to Build Self-Driving Cars
New York Times - 4 Jan 2018 by CADE METZ

PALO ALTO, Calif. — Before the car can drive without a human, one must first get behind the wheel.

As the driver accelerates, stops and turns on local streets, sensors on the car record what he sees and track how he responds. Then a team of engineers builds software that can learn how to behave from that data.

The software is installed in the car, and it can drive on its own. In the end, the car mimics choices made by the human driver.

This is how things work at Aurora Innovation, a start-up founded by three veterans of autonomous vehicle research, including Chris Urmson, who previously led the self-driving car project at Google.

The company’s methods are part of a change sweeping across the world of self-driving cars: a kind of so-called machine learning technology that promises a chance for little companies like Aurora to compete with the giants of both the tech and automotive industries. With it, researchers can build and improve autonomous vehicles at a far more rapid pace — one of the reasons Aurora believes it can close the gap on companies that have been working on self-driving technology for years.

On Thursday, the year-old start-up said that it had agreed to supply self-driving technology to the Volkswagen Group and Hyundai, two of the world’s largest car companies. Johann Jungwirth, the chief digital officer at the Volkswagen Group, which owns Audi, Porsche, and six other major automotive brands including the flagship VW brand, said the company has been working with Aurora for several months, with an eye toward developing both autonomous cars and driverless taxi services.

In 2010, when Mr. Urmson and his colleagues at Google launched the autonomous vehicle movement, writing the computer code to guide their vehicles was a painstaking, line-by-line effort. But in recent years, a type of computer algorithm called a deep neural network has come in from the edges of academia to reinvent the way many technologies are built, including autonomous vehicles.

These algorithms can learn tasks on their own by analyzing vast amounts of data. “It used to be that a real smart Ph.D. sat in a cube for six months, and they would hand-code a detector” that spotted objects on the road, Mr. Urmson said during a recent interview at Aurora’s offices. “Now, you gather the right kind of data and feed it to an algorithm, and a day later, you have something that works as well as that six months of work from the Ph.D.”

The Google self-driving car project first used the technique to detect pedestrians. Since then, it has applied the same method to many other parts of the car, including systems that predict what will happen on the road and plan a route forward. Now, the industry as a whole is moving in the same direction.

But this shift raises questions. It is still unclear how regulators and lawyers — not to mention the general public — will view these methods. Because neural networks learn from such large amounts of data, relying on hours or even days of calculations, they operate in ways that their human designers cannot necessarily anticipate or understand. There is no means of determining exactly why a machine reaches a particular decision.

“This is a big transition,” said Noah Goodhall, who explores regulatory and legal issues surrounding autonomous cars at the Virginia Transportation Research Council, an arm of the state Department of Transportation. “If you start using neural networks to control how a car moves and then it crashes, how do you explain why it crashed and why it won’t happen again?”

[ read the full article HERE ]