clock menu more-arrow no yes

Filed under:

Tesla's autopilot isn't special (but it's still cool)

New, 93 comments

An important non-revolution for self-driving cars

"We’ve been able to accelerate autopilot and bring it to market faster than originally anticipated," Elon Musk said to a gathered throng at Hawthorne Airport last week, the ex-Northrop facility in Southern California that both Tesla and sister company SpaceX call home. "It’s actually in production. In fact, every car coming off the line at Tesla at the factory has the autopilot hardware."

Musk has an outsized personality with a reputation for flair and hyperbole, and for the most part, he earns it: both Tesla and SpaceX have undertaken and delivered on profoundly large challenges that entire industries (and governments) have failed to overcome. And his preferred term of "autopilot" for the bundle of hardware and software installed on new Teslas isn’t inaccurate — the car can take control on highways, park itself in your garage at home, and intervene when it thinks you might be on a collision course.

But for as many things as Tesla’s autopilot mode does, the one thing it doesn’t do is change the game. The stereotypical vision of a car of the future tooling around your neighborhood with a driver comfortably asleep at the wheel (or missing altogether) isn’t any closer to reality than it was before.

Only incrementally more advanced than what's already on the road

While it’s true that Musk introduced the most advanced commercial autonomous driving system yet, it’s only incrementally the most advanced when compared to what’s already on the road. The sensor package that all Model S vehicle rolling off the assembly line now have installed bears much resemblance to what various automakers have been offering for years: radar, forward-facing cameras, and ultrasonic sensors are new to Tesla, but they aren’t new to the industry at large. For instance, the Lexus HS250h that I used to drive, manufactured in 2010, could effectively drive itself on highways using its lane-keep assist and dynamic cruise control — between those two features, the car had full control over steering, acceleration, and braking. Like Tesla, it used a combination of radar and cameras.

Of course, sensors, software, and the legal landscape have all matured since then. There’s little question that Tesla’s system is the most advanced and the most integrated to get a production commitment, but others like Mercedes-Benz’s Steer Assist — which is available on the 2014 S Class — come awfully close. Some BMWs can already read speed limit signs, as the Model S can now do. And while there isn’t another car on the market that will park itself and pick you up unattended, Audi has demonstrated it. Meanwhile, a startup is preparing to ship automated driving as an add-on for Audis that are already on the road.

Automakers from Volkswagen to Volvo are converging on approximately the same place: hands-off driving in easy, relatively predictable conditions — highways, mainly — in the next couple years, with full autonomy likely to come a few years after that. (NHTSA refers to these systems as Level 3 and Level 4, respectively.) Tesla’s in roughly the same place, it’s just pushing a little harder by promising a 2015 launch and a campaign of rolling software updates to make the system more advanced over time. And, of course, Tesla has the X-factor, Elon Musk.

The technology can’t really move any faster than that. Just ask Google, which has logged hundreds of thousands of miles testing self-driving systems over the past several years but hasn’t solved snow and heavy rain in its new car (and it’s equipped with a wider variety of sensors than Tesla is using). The challenges to making a car that can safely transport you from door to door — no matter where those doors are and what the prevailing conditions may be — still loom large.

There will be many stepping stones along the way

But there will be many stepping stones along the way, each system a little more autonomous than the one before it. By 2017, you’ll see a full host of highly automated systems available on cars, and not just on expensive ones. Tesla’s lower-cost Model 3 will undoubtedly offer it, as will many others. By automotive standards, 2017 is just one mid-cycle refresh away. It’s the blink of an eye.

From there, Tesla runs into the same challenges facing everyone working on self-driving technology. The first, and perhaps the biggest, is a protracted policy battle: self-driving cars are finding friends in politics one state at a time, and only four states currently have laws on the books for provisional licensing of autonomous cars. The rules don’t rewrite themselves overnight, and there are still plenty of questions to be answered about how these cars will function in real-world situations and distribute liability. Even if fully driverless vehicles were ready today, it’d take years before they were broadly legal to use in all conditions nationwide.

"Tesla’s a Silicon Valley company," Musk noted in a recent CNNMoney interview about autonomous driving. "If we’re not the leader, shame on us." And yes, Tesla will be first to market with a particular set of autonomous features — but other automakers came before it, and more will follow. Each launch advances the self-driving conversation a quarter notch. Ultimately, it’s not going to be about who’s first — it’ll be about getting millions of driverless vehicles on roads large and small, safely, with nary an incident in sight.