I won't have the time for timestamps, but just check the first few minutes of some of his newer videos. It is very obvious. Like Alan said, hands are just easier.
Your brain is much better than Teslas'.
haha, except when they don't like fog, heavy rain, etc. I would welcome the return of radar for so many non-perfect days (ie. outside of sunny fair weather places).
I agree. I don't want a self driving car that drives as well as I, I want a self driving car that does a better job than I. I think Tesla has based their self driving technology on a myth, that is we only use sight to drive. I would argue we use all our senses in combination with sight.
In addition, we use a lot more detail such the position of the front wheels on the other vehicles, the head position and eyes of the other drivers and even the physical condition of the other vehicles ( dents on all four corners).
You raised my hopes but very limited, only 1 car each on Teslascope and Teslafi.
Didn’t the tweet from yesterday say it would go to employees first? There could be what we are seeing.
Shows 1 on teslafi in one area but 2 in another either way this should be rolling out wider by Friday if all goes well
I think I see where Alan is going, and I agree. Cameras are needed to drive, period. Without them, the car cannot drive on radar/lidar alone. If the cameras are obscured for any reason, the car cannot drive, even with radar. Radar will help in some cases, but if it's foggy, heavy rain and the cameras cannot see the traffic signal, or the speed limit sign, etc, then the car cannot drive. If you're expecting the car to drive in heavy fog and heavy rain with radar, how is it going to see the red light ahead? How is it going to see the lane markers to stay centered, or the road signs to keep the correct speed?
If they can’t reliably stop and measure distances with the current vision and compute system, they have much much bigger problems than a lack of radar.
Here's one along those lines. I was going to the airport in Medford, OR to pick up visitors a few days ago. Here's a photo of the map and related step-by-step directions:
Yep, I bet most of us experienced something similar.
Radar (and ultrasound and lidar) are good supplements to help with limitations of cameras and image processing. They can do some things far easier and better than a vision based system can you are right - ‘regular’ cameras are absolutely required.
The FSD team said (at first AI day?) that radar was too noisy/ too low res for proper fusion with vision. Imagine a passenger who yells at every other overpass " Brake! Hard! There's a car right in front of us!" There was too much disagreement between the sensors and as vision improved, radar became a distraction instead of a supplement. A better radar may help in low visibility scenarios to become superhuman. But to get on par, proper vision plus "memory" should do. They already have short term memory (occluded obstacles are remembered) and map data plays the role of long term memory. Only, it's often wrong. I believe they still use OSM (Open Street Map) for navigation. As it's open source, everyone can contribute, e.g. here.
