Yesterday George “Geohot” Hotz announced his self-driving kit on stage of Techruch Disrupt in San Francisco. Those of you who read the blog for a while now, know that I recently have been quite interested in artificial intelligence and automated driving. So why not look into a kit that has the potential to turn every car into a self-driving one. Is that even possible?.
During the presentation Hotz was super cocky and I got offended by how aggressive he is. He blamed others to be sell-outs, which I agree, but than again he hasn’t done anything yet that earns him more. Despite the fact that he is a mediocre presenter, those are all personal impressions. What is more important is what he actually announced. After consulting a friend and despite my first impression of Hotz himself, I got more into it and I start to understand the impact comma.ai could have.
During his keynote Hotz notes that his system is not able to make a fully self-driving car. It is more of an ADAS that is comparable to the autopilot a Tesla has. However, without the requirement of purchasing an entire new car. The kit includes a dash cam that is installed where the rear-view mirror is located. It combines the visual inputs and the sensorial infrastructure of the car to offer some advanced driving assistance to any possible car. At least this is the vision of the startup. At the current stage it is only able to support a small group of specific cars, since the hardware prerequisites within the car needs to be met.
According to Hotz the main USP is the “shippability” of the module. In comparison to all the other efforts in the market, the company is able to offer a real product to consumers. Comma.ai is able to deliver a module by the end of the year and would be able to ship its components at a price of $999 (plus a $24/month subscription fee).
Showing some parallels of the Microsoft vs. Apple feud in the 1990s. It is not the dash cam that is interesting when looking at comma.ai kit. It is the software. It basically combines the visual inputs of the camera with the sensorial inputs of the car. Through this, Hotz is able to apply his algorithms to create a relatively reliable advanced driving assistance system that enables lane-keeping and emergency breaking to theoretically a large amount of cars, as long as some base technologic elements are available in the car. Hence, if it works, he offers his technology to a wider part of the industry than most of the car manufacturers themselves.
If we want to establish smarter cities where vehicles are able to commute independently, it is necessary that cars, independently from the car’s brand, are able to communicate with each other. So far, most of the car manufacturers are working for themselves on those challenges. Like Microsoft made its operating system available for various computer manufacturers, I could totally see how a software company like comma.ai or HERE is able to be more disruptive than the mono brand efforts of the Teslas and Fords.
In the end it is necessary for pioneers in the industry to work together. They all share the same vision. We have to understand that a “we-against-the-rest”-mentality won’t achieve this vision. Don’t you think?
#QOTD: Would you buy such a kit if your car meets the prerequisites?