Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register
This site may earn commission on affiliate links.
Nice to have that information confirmed @verygreen that it does see cars but only shows those relevant to AP. Well done!

Compare the the backup camera image with the rest of cameras. They don't exactly match.

Was your backup camera still displaying the same frame the next time you tried to put in reverse? I had a similar situation when I first installed .46 and was about to back up. The backup camera showing an image from several hours ago. Had to reboot, everything been fine since.
 
  • Informative
Reactions: Ulmo
@verygreen @bswn1 The rear view camera has stuck in the past?

There's a thread about the rear-view camera issues (mine has been acting up since .46.8 too): Dangerous backup cam freeze?

Edit: Ah, @verygreen had responded on the next page. :)
There's a twist here.
The frame from the past is before I started to back out from that parking lot (or rather, probably soon after I pulled in judging by the light) , BUT the image on cid was working ok in that particular occurrence!
Moreover on that trip, I stopped at another place before getting on that highway and then backed out and it was also showing ok on the cid, yet APE camera thread gave up on that camera long ago, so there's some complex dynamic ongoing it seems.

The "green frame" one I actually did a multiframe snapshot and it's all 10 frames that are all zeroes so it took awhile for ape to realize the camera is feeding garbage before it stopped writing (and keep in mind it's 1 fps, so at least 10 seconds). Also green frame is from a trip where I got a "white" backup camera image on cid and rebooted cid with the two-finger salute.

It's good that there's some failsafes, but it's bad that they seem to take a while before triggering, I guess.

IMG_20171210_155052.jpg
 
Last edited:
The frame from the past is before I started to back out from that parking lot (or rather, probably soon after I pulled in judging by the light) , BUT the image on cid was working ok in that particular occurrence!
Moreover on that trip, I stopped at another place before getting on that highway and then backed out and it was also showing ok on the cid, yet APE camera thread gave up on that camera long ago, so there's some complex dynamic ongoing it seems.

The "green frame" one I actually did a multiframe snapshot and it's all 10 frames that are all zeroes so it took awhile for ape to realize the camera is feeding garbage before it stopped writing (and keep in mind it's 1 fps, so at least 10 seconds). Also green frame is from a trip where I got a "white" backup camera image on cid and rebooted cid with the two-finger salute.

It's good that there's some failsafes, but it's bad that they seem to take a while before triggering, I guess.

Thanks for the added insight!

It sounds to me there may be two possibly related issues there: the APE-CID rear-view camera passthrough acting up yet sometimes working, as well as APE capture of the rear-view camera acting up, sometimes independent of the passthrough working. Most of us wouldn't know if the latter is acting up since we can't see it...

I expect CID is working just fine in all these instances and the problem is what APE does to its signal in the middle.
 
I've been experimenting with ALC on .46 and it definitely is smoother and I did observe, anecdotally, that it remembered where the cars were in adjacent lanes (I'd been seeing that for a while). But its like @jimmy_d observed in another thread -- the car will refuse to change lanes for a period of time after a car passes and its recognized (there is some kind of cooldown trigger). So big thanks to @verygreen for revealing they are being ID'd but does the actual classification work? In both listed up thread they are "cars" which might have been true but we also know that's all AP2 displays. Can we verify with a truck or motorcycle as the lead car that the field remains cars (or does it internally classify correctly)?

Also, why does AP2 not reveal all the cars during a lane change (It soon dissolves the cars on the previous lane after you're in the "target" lane. It is useful to keep tracking and I've noticed ALC will sometimes lose the target lead car (it will disappear in the IC and car will accelerate to max speed and then reacquire lead and slow down). Its not 100% smooth though its tolerance for lane changing in tight quarters has improved considerably but I'm still not 100% comfortable since it isn't 100% at tracking cars.
 
  • Like
Reactions: TrafficEng
In both listed up thread they are "cars" which might have been true but we also know that's all AP2 displays. Can we verify with a truck or motorcycle as the lead car that the field remains cars (or does it internally classify correctly)?

In the context of ALC, does it matter if they are cars/trucks/bikes, or is it just a driver confidence aid?

Functionally there is no difference - AP will maintain the preset distance, and will give the vehicle a full "lane" of width.
 
In the context of ALC, does it matter if they are cars/trucks/bikes, or is it just a driver confidence aid?

Functionally there is no difference - AP will maintain the preset distance, and will give the vehicle a full "lane" of width.

Each vehicle type behaves differently and the rules of the road are different. It does matter if you want a good driving policy. It also affects where is acceptable to drive. For example, trucks are wider and require more distance when following, passing and actually overtaking. Further, motorcycles can traverse lanes differently, including between lane lines. They can stagger in groups within the same lane. This isn't merely window dressing to reassure drivers...
 
Each vehicle type behaves differently and the rules of the road are different. It does matter if you want a good driving policy. It also affects where is acceptable to drive. For example, trucks are wider and require more distance when following, passing and actually overtaking. Further, motorcycles can traverse lanes differently, including between lane lines. They can stagger in groups within the same lane. This isn't merely window dressing to reassure drivers...

It matters for EAP (assuming EAP = auto-overtaking) and FSD, but for simple ALC? I'm not convinced.
 
Got another interesting glitch today - color mismatch between real world and what the camera reports. I guess we need to count our blessings the forward-looking cameras are not color (though I guess if the sensor is confused about which pixels are where it would still be equally bad).

backup5.jpg


IMG_20171212_162548-w2.jpg


And while we are at it, this is how it looks like when the sun is behind you to the backup cam. I guess it is never going to be any sort of a primary sensor for AP for all these reasons.

backup6.jpg