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

Tesla vision only - no radar from 2022 Q2 cars

This site may earn commission on affiliate links.

We know Tesla are going all in on "Tesla vision" and looks like Europe is next if this article is believed. Looks like my new MY would be one of those new non radar ones. Personally I think it's the wrong route to take as sure having more sensors and things to monitor the surroundings is a good thing and not a bad. What does everyone else think?
 
My experience is, the only time my M3 did a serious braking to an almost halt, was when a vehicle three cars in front suddenly braked in the outer lane of a dual carriageway. I hadn't seen it, although I was a safe distance behind the car immediately in front (follow distance set to 5), the queue of cars ahead were rendered on the screen and it was clear that somehow the radar had picked up the sharply slowing vehicle (marked it in red and gave a warning bleep) as it wasn't visible in line of sight due to the cars ahead, to us in the cabin at least. I don't know how far ahead the radar can see, but on this occasion, I was impressed.

Don't think there's anything you can do about it, other than buy an older car if you really want the radar option.
 
having more sensors and things to monitor the surroundings is a good thing and not a bad

Provided all the sensors agree. If they don't then ditching the one(s) that are "more often wrong" allows the processing to focus on / do more with the better quality data.

the only time my M3 did a serious braking to an almost halt, was when a vehicle three cars in front suddenly braked in the outer lane of a dual carriageway.

I remember when Tesla first added this ("Look under the car in front, to see what the one in front of that is doing") there was a dashcam from a Tesla that jumped on the brakes. The brake lights of the car in front never came on ... it just slammed into the car in front of that, which had presumably made an emergency stop and the following driver not concentrating / too close / didn't react. Impressive video ...

I doubt I would have reacted in that situation - i.e. absent the brake lights of car in front it will either be sudden closing distance (car in front braking, brake lights not working) or Wham! as it hits something - if the car in front is coming to a dead stop that needs quite a bit of react-and-slow distance of course, so maximising that by car detecting within a few milliseconds that either closing-distance is rapidly decreasing, or car several in front suddenly slowing, is worth having.

Dunno how well vision works as a replacement for radar in that situation.
 
Provided all the sensors agree. If they don't then ditching the one(s) that are "more often wrong" allows the processing to focus on / do more with the better quality data.
Personally (based on no evidence) I would have guessed the radar would be more likely to give correct info then cameras which can become obscured/dazzled.

Only having one sensor to rely on is never a good thing...737 max anyone?
 
If Tesla didn't already have form for stripping parts off the car as cost saving measures and claiming that "data shows that it isn't used", like the passenger lumbar support, I'd have more confidence that removing radar was a determined act. As it is - I'm very sceptical.
 
based on no evidence

Google result are similarly full of opinion pieces. Someone who knows more about it than me can reply. My understanding is that Radar is "fuzzy" - "Sh1t! Something is coming at us". Vision's job is to figure out what it is, and whether it is a threat. Two eyes and a brain, that type of thing.

Only having one sensor to rely on is never a good thing...737 max anyone?

That wasn't multiple types of sensors was it? My recollection is that there was no redundancy, the MCAS software was found wanting and self-certification played a part in it being approved (and the "its not working" dashboard light was in the Options Pack)

Does Tesla have multiple Radar sensors? (multiple camera and ultrasonics, but just one Radar - is that right?)
 
I my experience the radar on the Tesla isn't that good - or not processed very well. Two things that happen regularly to me:
  1. Pedestrian or cyclist crosses the road a safe distance in front of the vehicle. Five seconds after they are back on the side of the road the car slams on the brakes.
  2. On a two-lane road with a slight bend and the other lane has stationary traffic waiting to make a turn or similar: Visualisations correctly show the road curving and the traffic, but the car slams the brakes on while flashing the stationary vehicles red as if your planned/normal course of action would be to just drive straight into them rather than follow the lane.
Near where I live one of the major A roads has recently had a junction upgrade so there is now a flyover taking the main route over the top of the roundabout at the junction. Heading south the car is absolutely fine with the new road layout. Heading north it slams the brakes on as you pass over the roundabout as if it's also using the mapping data and it thinks you're about to barrel straight through the junction.

Like so much of Tesla's software it seems to be ambitious, but short of the mark.
 
Personally (based on no evidence) I would have guessed the radar would be more likely to give correct info then cameras which can become obscured/dazzled.

Only having one sensor to rely on is never a good thing...737 max anyone?
There's a video from the developers where they explain that the radar doesn't get things right in specific situations, For example when the car in front brakes harshly or when going under bridges. The radar has much lower resolution that the cameras and sometimes thinks the bridge is an object in the road. Tesla are able to get vision to do everything that radar can do e.g. depth perception, see through fog, so the decision was made to concentrate development on vision rather than split the development teams over 2 separate and sometimes conflicting systems.

Incidentally radar can get obscured too. Used to get warnings on my last BMW every time it snowed.
 
  • Like
Reactions: teaston
Hopefully Euro NCAP will undergo new safety ratings since it's a substantial update, then we will see if the safety assist score remains at least the same.

They don't have to, but I hope they do and Tesla can supply the cars if they are confident.

The MY has not been tested yet.
 
Don't think there's anything you can do about it, other than buy an older car if you really want the radar option.
This won't be correct.

It will be a software update which all cars will eventually receive whether they have radar or not. As a software developer, maintaining one less input is much preferred if it can be dropped safely. All vehicles whether they have radar hardware or not will eventually be vision only as it's less for the developers to maintain.
 
Funny you should mention that. I’ve sold a few cars via Motorway with no issue. Most recent one was a bit of a pain. Reducing estimate for very minor things. Getting the reserve as low as possible. Don’t think I will use them again.
 
This won't be correct.

It will be a software update which all cars will eventually receive whether they have radar or not. As a software developer, maintaining one less input is much preferred if it can be dropped safely. All vehicles whether they have radar hardware or not will eventually be vision only as it's less for the developers to maintain.
Agreed. There could be a scenario that lack of radar doesn’t work out for them and they retrofit though. But yes, all being well, radar if fitted will become redundant
 
Hopefully Euro NCAP will undergo new safety ratings since it's a substantial update, then we will see if the safety assist score remains at least the same.

They don't have to, but I hope they do and Tesla can supply the cars if they are confident.

The MY has not been tested yet.
Don’t know if I imagined it but haven’t the Tesla safety reports they do every quarter stopped reporting passive safety performance (ie the type of thing the radar would be used for historically?). You can have your own opinion on self driving and whether radar is needed for that, but if emergency safety braking performance etc have been compromised then that’s an issue. It’s hard to explain why they’d drop a significant stat unless the data was, possibly temporarily, impacted. Tesla have been here before, dropping something before the replacement was as good as what it was replacing.
 
It’s hard to explain why they’d drop a significant stat unless the data was, possibly temporarily, impacted

There were restrictions on the cars that first came out with no radar (AP max 75MPH, follow distance minimum #3, no Summon / Emergency Lane Departure, etc.). They were all reinstated with OTA some time later. Maybe that was within the current stats period? But otherwise I agree, looks fishy.

Tesla have been here before, dropping something before the replacement was as good as what it was replacing

Like the MobileEye AP fallout ... I got my first car with MobileEye, I would have been properly annoyed to be one of the first to get a car with the replacement Tesla AP - how long was it that the car had no AP? My recollection is "many months"
 
Like the MobileEye AP fallout ... I got my first car with MobileEye, I would have been properly annoyed to be one of the first to get a car with the replacement Tesla AP - how long was it that the car had no AP? My recollection is "many months"
I recall it was about 9 months to get an attempt at AP1 features and more like 3 years before they performed about the same. That’s the worry as they take several months to get something working, but it can take a lot longer to get it working as well as what went before.

They also dropped a basic rain sensor on the windscreen and that’s still a topic of ridicule.