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

Vision Only Limitations

This site may earn commission on affiliate links.
Thank you, but what if you drive manually to 85, does it still lock you afterwards?
Exactly like driving to 95 in normal cars. If autosteer is on AND you exceed the limit, you’ll get locked out of autosteer for the remaining of the drive.

P.S. it’s not just the obvious limitations of 80mph and follow distance of 2. The overall autopilot performance on my vision-only Model Y is absolutely horrible compared to my normal vision+radar Model 3. An order of magnitude more phantom braking, and significantly worse general wondering on autopilot.
 
Exactly like driving to 95 in normal cars. If autosteer is on AND you exceed the limit, you’ll get locked out of autosteer for the remaining of the drive.

P.S. it’s not just the obvious limitations of 80mph and follow distance of 2. The overall autopilot performance on my vision-only Model Y is absolutely horrible compared to my normal vision+radar Model 3. An order of magnitude more phantom braking, and significantly worse general wondering on autopilot.
Got it. Sure seems like a step back
 
Cars can be deadly. Auto steer and auto pilot are tools to make cars safer.

But the auto steer technology is not fully baked yet so removing hardware while still developing the auto steer technology, without showing a clear benefit, is just cost cutting.
I feel like I'm taking crazy pills. Tesla has openly shown the data from the radar sensors failing to keep lock on cars in front (especially on a curve or hill) and detecting overhead items. If you have to expect regular false positives and false negatives you have to find mitigation strategies to cherry pick data you can trust or simply eliminate the untrustworthy sensor.

First they tried ignoring high up things and then cars slammed into semi trailers, then they fought phantom braking and subsequent rear-end collisions from people hitting a Tesla phantom braking. Then they fought cars failing to break for cars up ahead already at a standstill because radar never locked. Ultimately, it was easier to take vision to a point that is was equivalent or superior to functional radar than it was to use vision to mitigate the false positives and false negatives.

As far as bad weather and night performance, radar is not actually that great in rain and fog; it massively increases the issues radar already has to the point that you can't trust it to guide the car in conditions the cameras can't operate in. For night, radar works well because it emits its own EM energy; cameras do that too with their EM emitters, they're called the head lights. On the topic of emitting energy, radar was a power draw too, so win-win removing it.
 
  • Disagree
Reactions: Daekwan and ahmadr
That's funny. A lot of people seem to be experts at computer vision and remote sensing around here :)
Yep, almost as many as those that are experts at reading “true intent” into for-profit corporate decisions. For all of the supposed “future” benefits or current “issues” with combo detection issues, one thing that’s not up for debate is not having radar is cheaper to make/produce than having it. And there’s is ZERO outsider ability to determine which motivating factor was strongest (immediate cost savings/parts availability vs. actual “technical” superiority). IMHO it’s extremely likely a current step backwards in current (and potentially long-term) capabilities made for near-term profit reasons. Expert or not, I’ve been involved in enough high-level product development corporate decisions over the decades to be able to spot spin when I see it…and there’s a lot of it going on around this topic.
 
Yep, almost as many as those that are experts at reading “true intent” into for-profit corporate decisions. For all of the supposed “future” benefits or current “issues” with combo detection issues, one thing that’s not up for debate is not having radar is cheaper to make/produce than having it. And there’s is ZERO outsider ability to determine which motivating factor was strongest (immediate cost savings/parts availability vs. actual “technical” superiority). IMHO it’s extremely likely a current step backwards in current (and potentially long-term) capabilities made for near-term profit reasons. Expert or not, I’ve been involved in enough high-level product development corporate decisions over the decades to be able to spot spin when I see it…and there’s a lot of it going on around this topic.
Correlation or causation?
 
Anyone have an idea when the Tesla Vision Max Speed will be increased to at least 85?

I sold my wife's 2018 M3 FSD and bought her a 2022 M3 Perf no FSD and noticed the AP max speed is limited to 80 mph. My wife thinks this is a limit of not having FSD but she is going to kill me that it is a limit of her new car. She is going to kill me!
 
Anyone have an idea when the Tesla Vision Max Speed will be increased to at least 85?
I have no idea. It's temporarily limited until it will catch up with the radar version in the future (with no known time estimate).

I notice that the radar range is only 160 meters or 524 feet while the camera range is at the advantage of longer range at 250 meters or 820 feet.

5Wn1A6E.jpg
 
I have no idea. It's temporarily limited until it will catch up with the radar version in the future (with no known time estimate).

Side note here. Tesla is using test vehicles with a LiDAR unit front center. Apparently it’s being used to check the cameras and create new models for the cameras to use.

This should provide for drastic improvements in a number of areas, including likely top speed.
 
Side note here. Tesla is using test vehicles with a LiDAR unit front center. Apparently it’s being used to check the cameras and create new models for the cameras to use.

This should provide for drastic improvements in a number of areas, including likely top speed.

You mean in May 2021 after radar was discontinued in North America Model 3 and Y?


E11MdiyWEAEVs8C



or in 2016 for the FSD demo video filming?

screen-shot-2016-07-07-at-10-57-41-am.png


It has been 5 years so I guess the radar-parity should be "soon".
 
Last edited:
I have no idea. It's temporarily limited until it will catch up with the radar version in the future (with no known time estimate).

I notice that the radar range is only 160 meters or 524 feet while the camera range is at the advantage of longer range at 250 meters or 820 feet.

5Wn1A6E.jpg
so what you get down to is resolution of the cameras and update rate. how well/quickly can it detect an objects rate of change in size?

radar directly measures this. cameras have to calculate it based on pixel changes from frame to frame.
 
Anyone have an idea when the Tesla Vision Max Speed will be increased to at least 85?

I sold my wife's 2018 M3 FSD and bought her a 2022 M3 Perf no FSD and noticed the AP max speed is limited to 80 mph. My wife thinks this is a limit of not having FSD but she is going to kill me that it is a limit of her new car. She is going to kill me!
Yes that’s correct. All vision My max ap is 80mph. Been waiting for them to increase speed for awhile. I had a 2021 my lr with radar previously and it went to 90mph.