"Tesla secured a series of approvals from California required for a promised robotaxi service.
The California Public Utilities Commission said it approved Tesla’s application for a transportation charter-party carrier permit, a license typically associated with chauffeur-operated services, allowing the company to own and control a fleet of vehicles and transport employees on pre-arranged trips.
The permit is a prerequisite for applying to operate an autonomous ride-hailing service in California, but a CPUC spokesperson said the current permit “does not authorize them to provide rides” in autonomous vehicles, and does not allow Tesla to operate a ride-hailing service to the public."
It helps, but humans function without LIDAR (though it can be helpful to them too). I am confident that it is technically possible to drive with just a camera, given sufficient image processing. Is our image processing in 2025 sufficient for the task? That I don’t know.
Have you seen the recent video where a tesla fell for a wile e. coyote wall?
I got you fam, I posted a link for the person you responded to.
The issue isn’t image processing, but the logic of these systems. The algorithms/models running autonomous vehicles aren’t smart or flexible enough. The current type of systems just aren’t advanced enough.
https://youtube.com/shorts/U1MigIJXJx8
Wile E. Coyote…
I mean I don’t like Tesla either and I agree that they should use Lidars but how is a Lidar supposed to solve this issue? It’s still a solid surface and the Lidar will still pick it up.
That’s the entire point of Mark’s test, Tesla doesn’t use LIDAR anymore, just plain cameras. LIDAR wouldn’t be fooled by such a painted wall, but apparently typical cameras can indeed be fooled.
It’s similar to facial recognition unlock, if it doesn’t use some form of 3D cameras or sensors, then just a simple photo of you is enough to unlock your phone.
Oh whoops, I misunderstood what the video was supposed to show. Yea that tracks.