Uber’s Shitty Month – Part Deux

It’s been barely a year that a series of turbulences at Uber had reached a climax. In February 2017 several self-induced disaster hit Uber, ranging from sexual harassment to dangerous driving situation of their self-driving cars, which ultimately lead to the ouster of Uber-CEO Travis Kalanick.

Since then a lawsuit around alleged IP-theft from Waymo were settled with the payment of $245 million dollar worth of shares. And now this fatal crash with a self-driving Uber-car.

The details that are know so far are indicating that Uber seems not to have learned much from the events from the past year, despite a new leadership team.

It starts with looking at the absolute numbers that highlight how far Uber is lagging behind its competitors such as Waymo, when we look at safety of its autonomous cars. The latest Californian Disengagement Report from 2017, which has to be submitted by all companies that tested the full year in California, shows disengagement rates for Waymo every 5,400 miles (9,000 kilometers). Mercedes at the bottom of the list has disengagements every 1.2 miles (2 kilometers).

Uber wasn’t required yet to report for 2017, but leaked numbers from march 2017 showed a disengagement every mile (1.6 kilometer2), and the latest reports from the New York Times now show one every 13 miles (20 kilometers). Arizona does not request a disengagement report by the companies. Beside Uber, also Waymo and GM Cruise are testing in Arizona.

The reason why Uber is testing in Arizona is based in Uber’s shitty month from 2017. Uber did not even apply for a California test license, and when the state authorities threatened Uber to revoke their operation license, Uber loaded the cars on semi trucks and trucked them to Arizona, where the local governor welcomed the company with open arms and little regulation.

Although shortly after those events a new CEO was brought on the helm of the company, new company structures and processes introduced to change the aggressive and destructive company culture, and Uber finally applied for a test license in California, the details known so far from the fatal crash show that not much progress has been made. Uber is its own worst enemy, and that harms the public and all other companies developing self-driving technology.

The video that was published by Tempe police may show an exterior view with the 49-year-old victim suddenly coming out of the shadow, but the interior video shows a safety driver being distracted and not paying attention to traffic.

Experts find it surprising that the car’s sensors did not detect the woman and stop the vehicle. According to insider reports, the Lidar on the Uber car was turned off to test camera vision. That fact alone is horrifying. To turn off the Lidar and not have it running as backup system during such tests to step in in exactly such cases, seem to literally ask for such a crash to happen.

The visibility during that traffic situation on the video from the Uber camera is contrasted by videos from other tests that show how clear the view should have been.

Traffic safety authorities investigating the crash did some tests with the car at the crash scene. There is no mention of a sudden appearance of the victim from dark shadows, the visibility is good.

Uber’s strategy to use all means and forces to press forward with its efforts has gotten a big pushback. For the time being all further tests are stopped. The goal to start a commercial robotaxi fleet by the end of this year moved in the distant future. And with that moves the question into the center, whether Uber is morally and technically suited to develop such a technology.

The technology itself and developing it is not in question, but whether we want companies with behaviors such as Uber to work on it. Judging from today’s fact we can only come to the conclusion that Uber should not be given the permit.

This article has also been published in German.