The exponential progression of the autonomous car
The autonomous car is living in a hype. There is an explosion of news and commentaries around it. It’s normal, the simple fact of imagining you traveling in a car without a driver is, at least, very flashy. And it is also easy to imagine the enormous impact in so many aspects of the day-to-day of all of us.
Everyone with whom I spoke about the subject is positioned or in those who think that it will be sooner than we think, or in those who think it will be available after than we think. But there are very few who dare to specify when it is what we think. I believe the progress of the autonomous car is taking an exponential curve, by the hand of artificial intelligence. So the idea is to make a quick retrospective of the milestones that I consider most remarkable to be able to forecast when it is going to be a reality.
Tesla and Google already have the software required for level 5 autonomous cars
This reality also has nuances. I will consider three deadlines, the first is when the technology is ready; The second when it is marketable; And the third when it has a critical mass within the vehicle park to impact the cities.
June 1953. Isaac Asimov, in the short story “Sally”, talks about cars with positronic brains. It is relevant because science fiction writers often describe the vision that will inspire scientists and engineers. Simply that time from that vision to the execution is becoming much shorter.
October 2010. Google publishes a post explaining the work they are doing around the autonomous car.
July 2012. The state of Nevada gives Google the first license to test on its roads an autonomous car
October 2015. Tesla updates over the air their cars to incorporate Autopilot. This is possible because a year earlier, in October 2014, they began equipping their S model with the necessary hardware. It is a controversial name because it is level 3, so it requires the attention of the driver. It only works on the highway, but there are times when you need the driver to take control. I had a chance to try it and the problem comes from being in the driver seat you quickly have a false sense of security and you lose your attention to the road.
January 2016. Mercedes presents in the past CES a prototype of an autonomous car, the F 015, in which the seats can be put in front. What does not clarify Mercedes is when they plan to market the new model. In the presentation they mention 2030, but it is not clear.
February 2016. The NHTSA considers as a driver the software of Google. It is an important step because there is a lot of regulations in which the driver is referred to. In the style “the driver should be aware of the circulation”, well it is Google’s artificial intelligence software that should be attentive.
August 2016. Ford announces a model of autonomous car, and this time it is concrete. It will be marketed as a service in 2021 and it is level 4.
September 2016. Only a month after the announcement of Ford, Uber does the same but with a working service in Pittsburgh. A few days before nuTonomy, a spinoff from MIT, did the same thing in Singapore, but I believe the Uber case is more remarkable.
October 2016. Otto, a truck company bought by Uber, makes its first delivery with an autonomous truck of level 4. It is very relevant because the big piece of the endangered jobs pie is in this type of transport.
October 2016. Tesla announces that it begins to manufacture all its cars with the necessary hardware for the level 5 autonomous driving. It is an advertisement that seems repeated of what they did 2 years ago, but that was for level 3. Step 1 year between they made the announcement that the hardware was available in the cars and they updated the software.
December 2016. Alphabet (Google), after 8 years turns the autonomous car research project into a group company, Waymo. The step is relevant because it is easy to speculate that the moment it becomes a company is because they plan to market.
Finally, after seeing where we came from, my forecast is that Tesla and Google already have the software required for level 5 autonomous cars and are in the test phase — need to do many miles. They have the foresight to market that software within 1 year. And lastly, the renewal of the car park. In a country like Spain, of about 30 million vehicles with an average age of about 10 years and between 1 and 2 million new registrations per year depending on the economic cycle, it renews half the park in about 10 years. My forecast is that the adoption of the autonomous car is going to be accelerated and that cycle will be shortened to 5 years, mainly motivated by the change of ownership of the car to use a mobility service.
We’ll see.
(en español aquí)
