Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Current HW2 Autopilot using 2 of 8 cameras * Testing Inside *

This site may earn commission on affiliate links.
Curious as to how many of the camera on the new Model S are actually being used at the moment so I decided to do some testing today.

I started by first covered in the side fender cameras. This has zero effect on Autopilot and it worked as if they were not covered.

I then covered the side B pillar cameras. Again, zero effect on Autopilot.

I then covered the windshield drivers side camera. Again no effect and no error warning.

Next I covered the windshield passenger side camera and finally started to see an effect. The Autopilot steering wheel icon kept appearing and disappearing. I was able to engage Autosteer but it was pretty erratic.

I finally covered all cameras and Autosteer would not work at all, as expected. The odd thing was that the visual lanes would still pop in and out of view on the display but the AP steering wheel icon never appeared.







 
Great test. I'm actually surprised it is using two of the three front cameras. Are you sure the tape on the right side wasn't infringing on the center camera's field of view? I'd be surprised if it is using more than one.

I came to say the same thing. Considering the center camera is a wide field of vision lens, I would not be surprised if those two pieces of tape were actually interfering with just the center camera.

I'd bet right now the system is only using one camera.
 
  • Like
Reactions: Vitold
I came to say the same thing. Considering the center camera is a wide field of vision lens, I would not be surprised if those two pieces of tape were actually interfering with just the center camera.

I'd bet right now the system is only using one camera.

That is most likely true. Guess that gives us hope that it should be much improved once they start using more cameras.
 
That is most likely true. Guess that gives us hope that it should be much improved once they start using more cameras.
I completely agree. I would go as far as to suggest that certain things like road border detection and cut-in detection are going to be implemented with multiple cameras in mind for the long run, and unlike EyeQ3, Tesla might not be spending any wasted effort hacking that onto a 1 camera solution.

Perhaps they are still awaiting a robust camera alignment fix and once multiple cameras are activated, the performance will be night and day.
 
I finally covered all cameras and Autosteer would not work at all, as expected. The odd thing was that the visual lanes would still pop in and out of view on the display but the AP steering wheel icon never appeared.

HW1 does something similar if the camera's view is occluded, say by lens flare or by interior condensation.

The way I interpret it, those phantom lanes lines something like the hallucinations or tricks of the eye that you or I might get in fog or low light, or in an Isolation tank - Wikipedia. The AP system works pretty hard at finding lane lines. If it looks hard enough it'll find them, even on a blank wall or the inside of a piece of masking tape. But it also had a notion of "confidence", measuring how likely it is that a perceived lane line is real. In cases like this its confidence is low, so it doesn't enable AP.
 
  • Like
Reactions: BeesKnees
Fantastic job @HX_Guy - I've been waiting a long time for this!

Your observations make sense: I think Tesla is using a monocam setup right now, as they have about 1,5 years experience with the MobilEye rig.

What's new for Teslas Autopilot-team, is of course the new camera (lens), the new camera chip, and the new ADAS-processor (PX2). I think Tesla right now is just trying to get a 1-camera configuration to work properly and getting it to parity with HW1.
 
  • Like
Reactions: Lee_B
If only the FishEye-camera is at work in all AP2.0-cars right now, it would mean that your camera *range* is max. 60 metres, am I wrong?
tesla.com/autopilot

Not sure if it's possible to test this. One would have to try to keep a distance of 60+ metres or so and check if your Instrument Cluster animates the car ahead of you. I'm terrible at eye measuring and haven't got an AP2-car, so I don't know if it's the case
 
I been told you people from day one that they were only using 1 camera to replicate ap1. but you ppl refuse to listen.
secondly the camera they are using is the main camera not the wide.


Don't hold your breath that more cameras will be used.
The two rearward facing cameras will be used to change lanes only for example.
EAP is primarily based on one camera configuration. In the conference Elon said they use the user redundant front cameras. (Main and Narrow).


The real FSD development and EAP development are two seperate teams.
and when i say real FSD development, i don't mean the EAP dev team that will start working on partial FSD features when EAP is done.
 
Last edited:
  • Like
Reactions: cdub
I'm not saying you're wrong, but I wouldn't come to the same conclusion as you did.

What you proved is that AP2 needs at least 1-2 cameras to work correctly. You don't know for a fact that they're not using the side cameras. They could be using them as supplement information to help AP2 when the main cameras miss something.

Again, you might be 100% correct, and likely are, but this test doesn't prove that.

=====

The other experiment, which would be interesting, is to leave all cameras uncovered, EXCEPT the wide angle center camera. See what it does?
Again, it probably wouldn't give you a definitive answer as to how many cameras are used, but it will give more info. What I suspect is that AP2 doesn't work, and what that implies is that for AP2.0 to work is that it needs that center camera (it forms the main image).