Hacker News Reader

Comment by rafaelero | Hacker News Reader

Tesla FSD may be bad, but no way it is that much worse than Cruise. Looks like they are inflating their data.

Cruise has cars driving out without a driver in San Francisco right now. Tesla is a world away from being able to do anything close to that.

I would not be surprised at all.

The biggest problem with FSD is bounding box detection as from previous videos it routinely fails to identify objects in the road correctly.

A problem that LiDAR (which Cruise uses) is especially good at.

That magnitude of a difference would be pretty noticeable and I haven't seen Cruiser users posting amazing reviews on YouTube in comparison to Tesla's. Maybe I am using an incorrect proxy, but it's hard to believe that Tesla would be so far off.

The difference is noticeable.

You only have to look at Cruise versus FSD videos to see how the latter struggles with reliable object detection.

Also this is about Cruise the robo-taxi company and not the GM SuperCruise technology which isn't what I would consider a peer to FSD.

Why do you say this?

Are you kidding? 40k km with no disengagement? This is nuts. I don't think we have reached that level yet. And if we did, I can't believe Tesla engineers would be so oblivious to that that they wouldn't update their technology. The likeliest explanation is that the article's chart is using some weird metric that is not comparable between automakers.

You've given no actual reasons you find this hard to believe, just running on 100% vibes. The data reported to regulators doesn't lie. Tesla really is that far behind the competition.

Reply on news.ycombinator.com