On Wednesday night, Tesla, the electric car company helmed by Elon Musk, announced that all new cars in production would be equipped with “the hardware needed for full self-driving capability.”
Now, this does not mean the cars will be able to drive themselves. Not yet, at least. What it means is that every new Tesla car produced — including the Model S, Model X, and forthcoming Model 3 — will come equipped with $8,000 worth of sensors, cameras, and computers in the hopes that one day, Tesla will perfect the software that allows cars to drive themselves. If that happens, the cars can be easily updated.
Tesla is, in effect, getting all of its new cars ready for the day when software catches up to hardware. It’s also implicitly branding itself as the self-driving car company long in advance of actually having self-driving cars. If you’re dying to get an autonomous vehicle one day but also need a car right now, then Tesla’s there for you. Or at least that’s the pitch.
In its announcement, Tesla said that every new vehicle would include:
- Eight surround cameras providing 360-degree visibility around the car at up to 250 meters of range.
- Twelve updated ultrasonic sensors that allow for detection of both hard and soft objects at nearly twice the distance of the prior system.
- A forward-facing radar with enhanced processing to provide data about the world on a redundant wavelength, “capable of seeing through heavy rain, fog, dust and even the car ahead.”
- “A new onboard computer with more than 40 times the computing power of the previous generation [that] runs the new Tesla-developed neural net for vision, sonar and radar processing software.”
In the very beginning, this equipment won’t do much apart from gather copious data. But over the next year, Tesla plans to test a more advanced version of its earlier (but apparently flawed) Autopilot program, enabling cars to automatically stay in their lanes or avoid traffic while in cruise control. Once it has perfected these capabilities, the company will send software updates over the air — and all Tesla cars on the road will have them instantly.
Further down the line, Tesla hopes to move to even more advanced self-driving capabilities, using the data collected by cars on the road, although in a press call Musk said “it’ll take us some time” to develop those. The company posted a video showing a Tesla driving itself in a parking lot:
The company hopes to demonstrate a car traveling all by itself from Los Angeles to New York, without any driver guidance, by the end of 2017.
Tesla is still a ways off from fully self-driving cars
That said, a nifty demonstration is different from actual deployment. The company noted that it would still need to “further calibrate the system using millions of miles of real-world driving” before the technology could be shared with consumers.
The timeline here is still fuzzy. It’s one thing to design a fully autonomous vehicle that can perform well in some situations, or even many situations. But for fully self-driving cars to become widespread, they need to be able to perform well in virtually all situations. That means not just reading signs and lane markings on highways. It also means figuring out what other drivers or pedestrians are doing, or responding to unexpected situations. And computers still get stumped far too often.
"There's a long ways to go in all of these areas," Edwin Olson of the University of Michigan told me earlier this year. "And reliability is the biggest challenge of all. Humans aren't perfect, but we're amazingly good drivers when you think about it, with 100 million miles driven for every fatality. The reality is that a robot system has to perform at least at that level, and getting all these weird interactions right can make the difference between a fatality every 100 million miles and a fatality every 1 million miles."
In the interim, to deal with the very toughest situations, many companies developing autonomous technology will likely settle on a compromise: partially self-driving cars that hand the controls back over to humans when the computer is unsure what to do. That’s basically what Tesla’s Autopilot did, and it’s what the semi-autonomous cars that Uber is deploying in Pittsburgh do (i.e., there’s a driver behind the wheel at all times to take over in sticky spots, like when the car’s GPS goes awry).
In technical terms, we’re still at Level 2 or 3 automation — still a long ways from Level 5:
There’s still a lot of work left to do. Still, Tesla is optimistic enough that it’s already putting the requisite hardware in cars. At $8,000 per car, it’s a hefty bet.
- Tesla’s ludicrously ambitious product road map, explained
- 5 big challenges that self-driving cars still have to overcome
- The Department of Transportation’s comprehensive new policy on self-driving cars