Uber is arguing that its self-proclaimed self-driving cars really aren’t all that self-driving to justify its decision to forego an autonomous vehicles license in California.
Uber launched a pilot program for its self-driving cars in San Francisco last week, allowing users to ride in a self-driving car that’s monitored by a trained driver and engineer sitting up front.
Uber first launched the pilot program for its self-driving cars in Pittsburgh in September.
But unlike the Pittsburgh program, which went off without major hitches, Uber’s San Francisco launch had been underway only a few hours before the DMV threatened legal action against the company.
“If Uber does not confirm immediately that it will stop its launch and seek a testing permit, DMV will initiate legal action, including, but not limited to, seeking injunctive relief,” the DMV wrote in a Dec. 14 letter to Uber.
Anthony Levandowski, Uber’s Vice President of Advanced Technologies, is arguing that Uber doesn’t need to get the license because the cars do not meet the definition of autonomous vehicles.
“When we look at how the Tesla vehicles operate, we see us operating in the same exact manner,” Levandowski said on a Friday conference call with reporters.
But there are a few things wrong with using Tesla Autopilot as a defense.
- Business Insider/Corey Protin
Tesla Autopilot is at this juncture primarily intended for highway use. Tesla’s first generation Autopilot, which has gotten a few updates since its 2014 release, allows cars to automatically steer, change lanes, and merge onto highways. There are some other convenience features consumers can purchase for extra as well, like the ability for the car to park itself.
Tesla is upgrading its Autopilot system in new cars to add abilities like changing lanes without driver input, matching the car’s speed to traffic conditions for active cruise control, and merging off highways.
But as a whole, Tesla Autopilot really acts as an advanced cruise control with some useful parking abilities. Its capabilities fit it squarely as a Level 2 self-driving system.
Uber’s self-driving system, which I had the chance to experience behind the wheel, is intended for urban driving. Unlike Tesla’s Autopilot system, it can respond to traffic light changes, make left turns, and generally navigate an environment with more pedestrians. That added complexity really qualifies Uber’s cars under Level 3 autonomy.
Uber did not immediately return Business Insider’s request for comment on all this.
SAE International, a US-based association that develops standards for automobiles, has a six-tier system to define autonomous cars ranging from Level 0 to Level 5. The distinction between Level 2 and Level 3 is slight but important.
- SAE International
Level 2 is actually still considered a human-driver monitored system, where the system is responsible for steering, acceleration, and deceleration. The driver is technically responsible for monitoring anything that falls outside those three capabilities.
It’s why Tesla is extra cautious to not use any self-driving branding when referencing Autopilot.
In August, Tesla removed a Chinese term for “self-driving” from its China website following a non-fatal crash while Autopilot was activated. A Tesla spokesperson said at the time the change was made to address “discrepancies across languages” and was not prompted by the crash.
The move shows how Tesla consistently brands Autopilot as a Level 2 system. As part of that, Tesla wants its drivers to keep their hands on the wheel when Autopilot is activated. If a driver refuses to do so, a warning signal will go off asking him to return his or her hands to the wheel.
If a driver were to ignore that warning signal, Autopilot would disengage for the remainder of the trip.
Meanwhile, Uber’s cars act primarily as a Level 3 self-driving system, which is actually the first level where the car is qualified as an automated driving systems. That means the car is not only responsible for steering, acceleration, and deceleration, but for the monitoring of the driving environment as well.
The SAE International system notes that human supervision is still necessary at Level 3 – something Uber has complied with by sticking two people upfront. But because the car is technically responsible for more driving tasks and is, therefore, considered an automated driving system, it follows suit that Uber would need an autonomous vehicles license.
Additionally, there’s no interface in the self-driving Ubers to make it clear human supervision is necessary. When I was behind the wheel of a self-driving Uber, I kept my hands on my lap for the majority of the time and never once heard a warning signal asking me to return them to the wheel.
Uber’s self-driving Ford Fusions come with a toolbar indicating when a human needs to take over, but that’s the extent of what you can see:
- Business Insider/Danielle Muoio
Tesla cars, on the other hand, come with a digital interface behind the wheel to show you exactly what the car is seeing so you have real-time information about whether the car is struggling in certain environments before it even asks you to take over:
All of this is to say that Tesla both brands itself and packages its cars to emphasize that the driver needs to stay in control when Autopilot is activated, emphasizing it’s not a self-driving car.
Uber is lauding its cars as self-driving and then trying to back out of the responsibility inherent with that claim to avoid obtaining an autonomous vehicles permit in California, which costs a mere $150.
Smaller startups like Drive.ai and NextEV have gotten the California permit. Even Tesla has a California permit, which a spokesperson told Business Insider is for future testing, not Autopilot.
Uber should either get a license and own up to the fact that it’s trying to advance Level 3 autonomy or stop branding its cars as self-driving. At the moment, unfortunately, it’s trying to have things both ways.