Two unintended consequences of the autonomous car project
Google has an excellent PR machine. Among the several moonshots projects they endeavour was one into making cars self drive. At the beginning it was an experiment, however the PR machine -unintentionally- started making as if Goolge was going to disrupt their next industry: automobile. A situation that was quickly noticed by the establishment.
Consequence #1: This time, the establishment took notice
Silicon Valley has disrupted established industries they attacked. This time though, the car industry was not going to allow that to happen to them, not without a fight. After the raise in popularity of the self driving car from Google, car manufacturers quickly scrambled to build their own version of it. To the surprise of many, the car industry responded accordingly, and in short time, most car manufacturers have -now- a test version of their autonomous vehicle. This with the advantage of the establishment having the know-how to actually ship and deliver such a car, something Silicon Valley (aside from Tesla) simply doesn’t know… yet, because …
It's easy to engineer cars. It's hard to build a lot of them. (The cheap Ferrari, expensive Ford rule)
— Horace Dediu (@asymco) January 5, 2016
Consequence #2: At the beginning, autonomous cars might be more risky, but why?
Overhype is frequent in any new technology. In the case of autonomous cars, it’s on another level. Lets just mention on what few talk about, current self driving car limitations:
- Limited driving area: Simplifying a lot how these cars work, among the many things the car evaluate while driving is the environment around them. To make this evaluation, it compares the surrounding with an stored image (yes, image) of that location. Thus the car needs always network connection in order to download those images and compare them with the actual environment (in jargon: cloud connected all the time). By doing this, the car can identify the difference between the two and then identify possible obstacles. The image in itself is not the only parameter the car uses, but it’s a big one. On the same matter, that means that the car only knows a drive path that has been previously mapped, if the car enters -for example- on a parking space or an area the car manufacturer has not capture (for whatever reason), the car can’t go there on their own.
- As with the limited driving area, these cars also have a limited driving weather. As of the day of this writing, autonomous cars can’t drive while raining or snowing, they -obviously- get confused. This condition limits their initial deployment to places on which nice weather is frequent. Which is kind of ironic, the main advantage of the autonomous cars is that they’re safer than humans driving. So at the initial stage, humans will get used to them and -in the process- forget driving experience, only to be confronted with the fact that if it starts raining, the car will default to human control/driving. But now the human has lost experience because they drive less and less, while driving through rain requires more experience.
I’m certain that the car industry will overcome these limitations. However, even the most optimistic predictions say that, the technology needed to surpass these constraints will be economically feasible maybe by 2025. In the meantime, we’ll have to wait and see 🙂