A week and a half ago, we learned that Tesla is on a quest to hire more engineers to accelerate the development of its self-driving car technologies.
Tesla was already no slouch in the autonomous-vehicle world, having released its Autopilot feature into the wild just over a month ago.
We sampled Autopilot as soon as it hit the streets and were quite impressed, to put it mildly.
But evidently, Tesla CEO isn’t impressed enough. So he put out a call via Twitter for “hardcore” software engineers to take Autopilot from where it now — semi-autonomous driving under certain circumstances, such as on highways — to the mythical full autonomy of the “Minority Report” type: cars that can drive themselves 100% of the time.
This is a hugely important development for both Tesla and the auto industry. Regardless of how one feels about how Tesla got to where it is now and where it may wind up in the future, the company has provided tremendous leadership for startup automakers, electric vehicles, and autonomous driving.
On autonomous driving, Tesla was actually a bit late to the game, but it has caught up rapidly. It now sits squarely in middle of an industry consensus about self-driving vehicles. The view is that autonomy will evolve over the next 10 years, with major car makers gradually adding features to their fleets. Consumers will move slowly and steadily from the “super” cruise control features that are now appearing in cars to full autonomy.
A faster future
Tesla was part of that consensus. The more out-there ideas about self-driving cars were being explored by Google, which with its Google Car is trying to take the human driver completely out of the picture from the get go. The pace at which Tesla introduced Autopilot more or less mirrored what General Motors has been doing, just with a more aggressive go-to-market plan.
Until Autopilot actually wound up in Tesla vehicles.
Benjamin Zhang/Business Insider
It was clearly a burning bush moment for Musk. This is how it goes sometimes: you don’t see the future until some small aspect of it appears in the present.
Musk isn’t one to waste time, so little more than a month after he saw Autopilot in action, he was ready to double down on the technology. The fastest way to effect massive technological change in the early 21st century is through software. Tesla already has plenty of experience with this, effectively making its cars almost new in their capabilities through over-the-air software updates.
Model S sedan owners, for example, went to sleep with cars that drove dumb and woke up with cars that could drive themselves. Software did that.
Theoretically, Tesla vehicles are already equipped with enough cameras and sensors to drive themselves much of the time. It’s simply a matter of orchestrating the data. And that of course is the challenge that Musk has now made it his personal responsibility to tackle — the forthcoming Tesla self-driving Seal team is going to report directly to the CEO.
Everyone I’ve spoken with the auto industry thinks that full autonomy will probably happen, but they also know that it order to achieve full autonomy, massive amounts of data will have to be crunched and lots of money will have to be invested in upgraded the tech that currently allows cars to manage the cruise-control-plus autonomy we can now enjoy on a limited basis.
Musk hears that, sees how Tesla Autopilot has performed so far, and concludes that there must be a cheaper, faster solution.
What does he have to do to pursue that solution?
Well, he doesn’t need to construct a factory or develop an entirely new platform. He just needs to hire some programmers, set them down in front of laptops, and hover over them like a ruthless taskmaster. I hope these people know what they’re signing up for. Faster and cheaper doesn’t mean easier.
Fortunately, Tesla has a big advantage over, say, a new entrant to the automotive game that also wants to offer disruptive autopiloting: Tesla has a fleet of some 90,000 cars, roughly, on the road right now that it can use as a vast test platform.
Every Autopilot-enabled Tesla is already feeding data back to the mothership, providing a basis for tweaking the technology for future updates. Tesla’s vehicles are quite literally learning the roads that they drive on and are enriching the company’s overall mapping efforts. This is something of a secret weapon for Tesla autonomous-driving initiatives: its entire fleet can learn to drive itself.
The bet on software
If Musk is right about betting on software, then he could advance the timetable on full autonomy by half a decade. Big leaps in automotive technology are held back by hardware. In essence, cars are about as good as they can get, having been steadily improved over a century Future progress on the hardware side will be incremental, even for Tesla.
That leaves the software side for game-changing innovations.
Musk knows this. That’s why he isn’t wasting any time. And I wouldn’t bet against him. He’s used software to alter the banking system (PayPal), drastically improve the automobile (Tesla), and provide a much cheaper way to get rockets into space.
That’s an impressive track record. A month ago, electric cars didn’t naturally lead to autonomous driving. But now they do. Get ready to experience Tesla’s vision of the future, a whole lot sooner than expected.