Uber’s Self-Driving Future Looks Grim | Koeppel Direct

Uber’s Self-Driving Future Looks Grim


By now, the entire country has heard about the fatal accident in Tempe, Arizona, caused by an Uber-branded self-driving car. Elaine Herzberg, 49, was pushing a bike across the street when the Volvo XC90 struck her solidly, without so much as trying to break or swerve.

The accident is still under investigation, with experts trying to determine if the vehicle’s software was at fault. This is a huge blow to Uber and confidence in self-driving technology as a whole.

Uber, Waymo and Cruise Racing to the Finish

Uber, Waymo (Google’s car) and Cruise (General Motor’s project) have all been driving and collecting data, using a combination of self-driving technology and human operators who can record bugs in the programming and prevent accidents. However, Waymo and Cruise are largely operating in states that have reporting requirements, like California. Uber’s cars operating in Arizona have no reporting requirements, so it’s difficult to know for certain how they’re performing.

But initial reports indicate that Uber’s cars are not ready to be left alone at the wheel. Where Waymo’s cars can go an average of 5,600 miles without driver intervention, Uber has been struggling to meet a target of 13 miles before driver intervention is necessary. This is despite the Uber tech having been road-tested for a million miles by September 2017 and hitting two million just 100 days later.

Another Black Eye for Uber

Uber is walking on a thin line at this point.

Between their multiple scandals in recent years, legal troubles over intellectual property with Waymo and this recent serious accident, the future of Uber self-driving tech (or anything else) is uncertain. For the moment, tests have been put on hold in Arizona, Pittsburgh, San Francisco and Toronto. The Uber cars are grounded and there’s no telling when they’ll be road-ready again.

Some of the flaws in the software may stem from the way the company tested them. Initially cars had two drivers, one for safety and the other to mark data points. Then, in an effort to clock more miles, only one driver per car was permitted. That driver had to not only stay alert to problems on the road, but also record any useful data that could be used to make the car perform better.

All of this was a recipe for disaster with a car that wasn’t ready to drive by itself. This could be yet another nail in the Uber coffin, pending the outcome of the investigation.


Your browser is out of date. It has security vulnerabilities and may not display all features on this site and other sites.

Please update your browser using one of modern browsers (Google Chrome, Opera, Firefox, IE 10).