汽车 > 汽车联网移动性博客



Uber: We Don't Know What We Don't Know

by James Messeder | 2月 08, 2019

uber uber ii

Two pictures (above) tell a story. In one eight-year-old picture a Toyota Prius without a driver delivers a pizza in San Francisco. In a separate picture, a pedestrian has entered a crosswalk while a self-driving Volvo with a driver runs a red light.

Both pictures were made possible by Anthony Levandowski, formerly of Google and now with Uber and currently responsible for Uber's consumer and commercial vehicle automated driving development efforts. Levandowski got his start in the U.S. Defense Department's DARPA Challenge where he helped create the only two-wheeled entry; he went on to work on Google's StreetView and created two robotic technology start-ups, later acquired by Google.

The eight-year-old picture captures the so-called Pribot pizza delivery enabled by Levandowski's Anthony's Robots and facilitated by an OJ Simpson-esque police escort and closed intersections. Levandowski left Google after nine years to create yet another company, Otto, to bring automated driving to the trucking industry.

The trucking industry appealed to Levandowski, he told Business Insider last week, because of the simpler requirements for enabling automated driving primarily on highways where there are no traffic lights or pedestrians and because of the 50,000 unfilled openings for truck drivers. Uber acquired Otto, which landed Levandowski back in the world of consumer vehicle automation.

http://tinyurl.com/goz7h7e - "The man who invented the first self-driving motorcycle is leading Uber into the future" - Business Insider

It was Levandowski’s automated but human driven Uber that ran a red light in San Francisco last week at the outset of Uber’s testing in the state. The incident led to California regulators demanding that Uber stop its autonomous vehicle testing until it obtains an autonomous vehicle license like 20 other companies have. Levandowski has so far refused to bend to the demands of regulators to obtain an autonomous vehicle license because, he insists, the Uber vehicles are manned by human drivers just like non-autonomous vehicles.

What we don't know is how responsible the driver in the offending automated Uber vehicle was for violating the red light. We don't know exactly how automated Uber vehicles are handling intersections or even if they are capable of recognizing intersections and or recognizing the phases of traffic lights.

We don't know how automated Uber vehicles recognize or respond to pedestrians. We don't know if, like Daimler, an automated Uber vehicle presented with the dilemma of protecting the vehicle passengers or a pedestrian might opt to hit a pedestrian under the right circumstances.

We just don't know.

But, for Levandowski, all of this is a moot point because the presence of a human driver dictates that the driver will take full responsibility regardless of what occurs. (Uber reported that the driver implicated in the San Francisco incident has been fired.) The assumption here is that the driver is in control or capable of taking control at all times which suggests that the main purpose of the self-driving system is to gather data.

But automated driving is occurring at some times with the Uber vehicles which raises multiple questions such as:

Can the computer override the human driver?

Is it possible to determine from recorded data exactly whether the human driver or the computer was responsible for running the red light?

Can automated Uber vehicles recognize intersections and the phases of traffic signals?

Will automated Uber vehicles brake for pedestrians? At what speeds?

Will an automated Uber override the driver’s guidance to avoid hitting another car or a pedestrian?

The technology exists, from a company called Global Mobile Alert, to recognize intersections and the signal phase and timing of traffic signals and communicate that information to the driver. Might integration of that technology help prevent incidents like the one that has sparked controversy in San Francisco?

The two common threads between the two pictures are intersections and Anthony Levandowski. Levandowski left Google to focus on an automated driving scenario, trucking, that did not involve intersections. It looks like Levandowski has been returned to the crossroads with the industry facing a critical turning point on the path to automating driving. We’ll have to see who blinks first, Uber or California. Meanwhile, pedestrians in San Francisco will be wary.

Previous Post: Uber Nicht Über Alles | Next Post: BMW: From DriveOwn to DriveNow
Leave a comment