Hacker News Reader

Comment by bradknowles | Hacker News Reader

I wonder how happy GeoHotz is about this development?

Is comma.ai interesting to other companies at all? They have a system for integrating into existing cruise control, which Apple doesn't need (they can drop money to get cars made for them). They also have a vision system which is not very secret - throw lots of video footage into a NN - which should be possible to replicate relatively easily. Apart from aquihire, why would anyone buy comma.ai?

Is comma.ai's driving system at all realistic in terms of actual capability? I don't own a compatiable car (nor have the knowledge or courage) otherwise I would love to install and try to use it.

Depends what you expect from it. I think the official website describes it perfectly well: "openpilot improves your stock ACC and LKAS". It's a fancy, very active cruise control. From I've seen it seems to work really well.

A couple of years ago the founder was super confident that he would have a fully self driving car soon, and definitely before Tesla.

A couple of months ago I read that he decided the problem is basically unsolvable and a human will always need to be involved.

He's still right about being "before Tesla" as long as Tesla never figures it out either.

> A couple of months ago I read that he decided the problem is basically unsolvable and a human will always need to be involved.

Are you referring to the Engadget article? I understood him differently.

The copy on https://comma.ai/ has certainly changed tone from what it once was.

Now it is clearing pushing a driver-assist angle ("improves your stock ACC and LKAS," "copilot," "augment"). No mention of "self-driving," "autonomous," or the like.

Previously, it was clearly marketing a full-self-driving, no human involved system: "ghostriding for the masses," "software to make your car self driving."

It could be for liability reasons.

Reply on news.ycombinator.com