For years now we’ve ben hearing about developments in self-driving cars, from Google’s years showing off their own self-driving technology to the countless startups trying to solve the problem. Progress has been spectacular, especially over the past few years, as both the parts to manufacture a self-driving car and the software to power it have become cheaper and more accessible.
When it comes to autonomous vehicles, the real breakthroughs needed to make it possible are in the software. We’ve been building cars for over 100 years. Making a car go, stop, turn - those are problems that have already been solved, perfected, packaged, and shipped to consumers billions of times over. But having a car do all this on its own (and playing nice with the rest of the cars and humans on the road) is the real breakthrough needed.
The approach that every company is using involves ‘machine learning’, which pretty much means that overtime, the decisions/choices the software makes improves as it uses past data to improve itself.
Think of it this way. A car is travelling down a road at 40 miles an hour. At some point on the road, there’s a bump, so the car jumps a little bit as it probably should have gone over the bump at 20 miles an hour. The car can learn from that experience. Next time its lasers/cameras see a similar bump and it’s already going 40 miles an hour, it can correct itself and slow down to 20 miles an hour. Of course, this is a simple example and the intricacies of machine learning are more complex, but should make the next part a bit clearer.
According to
Google’s Self Driving Car website, their self-driving cars have driven over 2 million miles. That’s impressive, really. It’s also a lot of data that their software can learn from and improve itself through. Tesla on the other hand, has
over 140 million miles of data. Of course, those aren’t self-driven miles like Google’s, but it’s still 70x the data that Google’s self-driving cars have generated. That may help explain how Tesla is already preparing to offer consumers production-ready self-driving technology, which is of course the news here.
If you’re a new Pivoting subscriber, you may have not seen Issue #7 where I wrote about Tesla’s master plan. Definitely
give it a read if you’re interested in the long term ambition of Tesla and the genius of it all.
For now, here are some articles from the past week as Tesla showed off what it’s technology can do, how much it will cost (~$8000) and when consumers will get it.