Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

The level 2 vs 4 thing is a technicality and fairly misleading, Teslas can already operate in a much wider range of roads and conditions. Waymo is still on guardrails. Mercedes touted "level 4" but you had to be following another car, going slow, etc etc. General autonomy is what matters.

I trust Karpathy when it comes to lidar vs vision. Do you shoot lasers out of your head to drive?



LIDAR sensors are getting cheaper faster than camera-based autonomous driving software is improving. I predict that in a few years, regular luxury cars that are still mostly human driven will come with LIDAR for collision avoidance and improved driver situational awareness. Just like they already use RADAR for the same purpose today.


LIDAR is not some inherently superior solution. And more importantly, once you get away from hardcoded self-driving solutions and go deep on machine learning (the only viable way to reach L5), it can add more noise than signal if using sensor fusion.

Anyway, time will tell. I think in 2 years or less, Tesla will be absolutely dominant in autonomous ridesharing in the US.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: