The American start-up wanted to turn any car into a standalone vehicle with a connected box, but the US authorities showered his hopes. CEO returns on his turn to an open source solution for developers.
George Hotz. Our goal with Comma.ai is to develop the best autonomous driving system. If we do not market anything that can make your vehicle directly autonomous, programmers can use our hardware and software, including open source software, to experiment with autonomous driving. We employ 12 people and raised $ 3.1 million in 2016 from the Andreessen-Horowitz fund.
In concrete terms, what do you market?
We are selling a dashcam called EON (dashboard camera, ed) that allows to film and record driving data and on which it is possible to install software like OpenPilot, ChiffR, or ChiffRplus and even apps like than Waze or Spotify. These last two software allow to save the data from the dashcam. ChiffR is very easy to use, ChiffRplus requires a more advanced level while OpenPilot is more designed for programmers. We also sell two devices called Panda and Giraffe, which connect easily to a vehicle and, like sensors, can collect more data. Finally, another software called Cabana is intended to facilitate the interpretation of this data manna.
Which vehicles is your technology compatible with?
We want our technology to be used in all vehicles. It currently works with six of the ten best-selling car models in the United States, representing several million vehicles sold each year. At the moment the Opensource Openpilot software (which allows to activate the autonomous driving by acting on the various sensors of a vehicle, ed) is only compatible with the models of marks Honda and Toyota, very popular in the United States because of their reliability. If we chose these two brands, it was because they gave us access to a large part of the market. Conversely, Tesla is a little iOS of the autonomous car: their platform, Autopilot, was developed solely for their vehicles. We want to be Android.
In 2016, you had canceled the marketing of your first product, the case Comma One – a kit for autonomous vehicle – following a letter from the US authorities. Is this new strategy based on open source software a way to bring you into compliance with US law?
We were not stopped because of the legislator. The Comma.One has never been thought for a consumer experience. People misinterpreted what we wanted to do.
The current system still seems complex for the lambda driver who wants to make his vehicle autonomous. Do you plan to offer a more consumer-oriented experience in the future?
Maybe we’ll do it if no one else does it, but that’s not what interests me. My goal is mainly to solve the technological problem of autonomous driving. Other companies will find a way to pack our technology to create a consumer-oriented offering. In fact, two young companies have already tried and I think we should see more this year. These companies do not work for us and this is precisely the interest of open source: those who wish to use our technology to offer a consumer offer around autonomous driving are free to do so.
How do you intend to distribute your technology? Is your goal to offer it to car manufacturers?
I have no intention of selling our technology to car manufacturers. The idea would be to distribute it via dealers or aftermarket. Our devices are now sold directly through our site. To date, we have sold over two thousand Pandas. Last November, these 2000 drivers had traveled more than 2 million miles. This driving data was measured on all OpenPilot software, ChiffR and ChiffRplus. Note that only a few hundred of these drivers used OpenPilot to experiment with autonomous driving.
Why exclude a commercial relationship with builders?
Because it’s impossible to talk to these companies. It’s not just a technological issue but also a culture issue. Even if some of these automakers like to repeat that ‘autonomous vehicles represent the future’ they do not understand what this really means. For example, look at their navigation systems: why do not these manufacturers develop better GPS systems? Why do not they integrate services like Google Maps or Waze as most of their customers would like? For them, it is only a simple checkbox in the list of features they must integrate. Tesla is the only manufacturer to offer a browser with a large screen and integrates Google Maps. In my opinion, this company is also the only one, with Volvo, to really project itself into the future.
Is the company profitable? Will you need to do a new round?
We are very close to profitability. For now, all of our revenues come from sales of our material on the Web, but we may be able to monetize our technology through a subscription in the future. Finally, if the opportunity to raise funds with very attractive conditions comes, I would study, but I do not need to hire many engineers. I remain convinced that 10 very talented engineers can accomplish more than 100 average engineers. This is why our recruitment process is extremely selective.
What are your plans for the future?
For now, our only goal is to improve our software. My goal is to solve the technological problem of autonomous driving. Concretely, as long as cars do not drive at least twice as well as humans, I will not launch any other challenge. But we still have some way to go to get there.
George Hotz is the founder and CEO of Comma.ai, a company whose goal is to allow the installation of autonomous driving on each vehicle. Before that, he was known for several ‘hacks’. He became known for the first time at age 17 when he successfully unlocked the first iPhone. He will talk again about him by succeeding in hacking the PS3.