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

Radarless Driving Report? Anyone?

This site may earn commission on affiliate links.
There was a report from someone who took delivery during the recent rain in the northeast, and AP/TACC had them take over 10 times in a 30 mile turnpike drive due to poor visibility from the rain /mist kicked up by cars in front of the Tesla.


hopefully in future sw they overcome this, but I’m worried there’s only so much that passive photon sensors can do for safety to see through a spray cloud, unlike radar.

this seems to me as a case of the long 9s/ long tail in CA, that hasn’t been trained in the neural net yet. I’d love to see them train with lidar ontop in heavy rain and snow to have it learn how to gauge distance without seeing the full object at all times. It’s not an easy problem, it involves NN knowing what that object looked/ normally looks like while only seeing in some cases the top of it, having it disappear from view at times in the mist, and object permanence: extrapolating the best guess of where it still is based on past trajectory and visual cues like suddenly redder mist or snow from brake lights. Also something like an empty uhaul trailer towed behind a normal car would be invisible to cameras at times, and it would need to “remember” that that vehicle needs additional following distance
 
Last edited:
There was a report from someone who took delivery during the recent rain in the northeast, and AP/TACC had him take over 10 times in a 30 min drive due to poor visibility from the rain kicked up by cars in front of the Tesla.


hopefully in future sw they overcome this, but I’m worried there’s only so much that passive photon sensors can do for safety to see through a spray cloud, unlike radar.
For comparison, I drove from NJ to Maine yesterday in on and off heavy rain and very low visibility. Used AP throughout the trip with zero AP disengagement. I took delivery in June 2020 and my car has radar.
 
For comparison, I drove from NJ to Maine yesterday in on and off heavy rain and very low visibility. Used AP throughout the trip with zero AP disengagement. I took delivery in June 2020 and my car has radar.
This is impressive/ sad to me. It was raining hard at times yesterday. On 95 I slowed to 60 or 55 myself just to ensure I had visibility. I have a modelY reservation, but would rather a car that enhances the safety of what I can see, not equal to…
 
As I work with AI/ML models every day, including computer vision, this is only proving my point. I just don't see how AP works without radar (or LiDAR) in bad weather, period. Although I'm happy my June 2020 build has radar, I'm sure Tesla will abandon the code for it, if the future for them is radar-free. The shame is that they will most likely regain their NHTSA/IIHS status, as those tests occur in excellent conditions.

I figure that enough posts like above, and unfortunately accidents, will occur and this all could change Tesla's direction, but this is where I'm actually starting to consider changing brands. Do we really need a yoke steering wheel, no shifter/wiper/turn signal stalks, or no radar? No. I'm all-in for innovation, but I feel they are just trying to be different for the sake of being different now. The Mach-E is really an amazing V1 EV, and I'm sure Ford will sell a ton of F150 Lightnings. I was thinking that VW would be the next dominant EV maker outside of China, but now I'm not sure...

For now, I will be absolutely holding off on any future software updates until this is figured out. As news like this spreads, I don't think I'll be alone. Tesla can't force it's customer's to downgrade their driver assist technologies, and this almost certainly will result in a class lawsuit.
 
Last edited:
  • Like
Reactions: dyau
As I work with AI/ML models every day, including computer vision, this is only proving my point. I just don't see how AP works without radar (or LiDAR) in bad weather, period. Although I'm happy my June 2020 build has radar, I'm sure Tesla will abandon the code for it, if the future for them is radar-free. The shame is that they will most likely regain their NHTSA/IIHS status, as those tests occur in excellent conditions.
This is my concern that the system is nerfed for when you need it the most. I was also concerned that the NHTSA/IIHS testing wouldn't encompass any inclement weather. I would suggest writing Consumer Reports and bring this up as an issue with them. Whether you like them or not, they can be a useful resource to help get safety issues remediated more quickly.

It seems likely when there were M3 braking issues in 2018 that the visibility CR brought to this by removing it from their recommended list and being vocal got Elon's attention. I think the same here would be a useful tactic.
 
I’m thinking about delaying my MYP delivery now. I really don’t want to be a beta tester for Tesla’s vision only AP, especially if the main problem was a shortage of radar parts. I’ve tried auto high beam before and quickly disabled it. It’s very distracting to have them flicker on and off at night.

Hopefully everything should be straightened out by Q3.
 
I am most interested in how it does in crappy weather and poor visibility.

For comparison, I drove from NJ to Maine yesterday in on and off heavy rain and very low visibility. Used AP throughout the trip with zero AP disengagement. I took delivery in June 2020 and my car has radar.

My experience mirrors that of @vinny84. I took delivery of my MY yesterday - a crappy day...rain, low light and poor visibility, etc. Drove home on two different interstates. After calibration (which maybe took 10 minutes?) I engaged AP and used it the entire way save for the interchange. I disengaged manually only once besides then because I was mucking around with settings.
 
  • Like
Reactions: DanDi58 and Zhe Wiz
Another unfortunate datapoint from someone who has both radar and non radar MY


This report references the radar being responsible for phantom breaking and this is the first I’ve heard of it. If so, I would think that would be easy to fix because it is a multi path problem and other radar applications have to deal with the same issues.
 
This is my concern that the system is nerfed for when you need it the most. I was also concerned that the NHTSA/IIHS testing wouldn't encompass any inclement weather. I would suggest writing Consumer Reports and bring this up as an issue with them. Whether you like them or not, they can be a useful resource to help get safety issues remediated more quickly.

It seems likely when there were M3 braking issues in 2018 that the visibility CR brought to this by removing it from their recommended list and being vocal got Elon's attention. I think the same here would be a useful tactic.
I haven’t checked consumer reports but I could’ve sworn I saw a post from this week that a national safety certification was removed from the m3 and mY after this radar development, and subsequently consumer reports removed their top pick and safety recommendation for those models.

At the time the consensus was consumer reports was spreading fud and biased towards legacy automakers and the establishment, but I’m beginning to think they were just being realistic .
 
As I work with AI/ML models every day, including computer vision, this is only proving my point. I just don't see how AP works without radar (or LiDAR) in bad weather, period. Although I'm happy my June 2020 build has radar, I'm sure Tesla will abandon the code for it, if the future for them is radar-free. The shame is that they will most likely regain their NHTSA/IIHS status, as those tests occur in excellent conditions.

I figure that enough posts like above, and unfortunately accidents, will occur and this all could change Tesla's direction, but this is where I'm actually starting to consider changing brands. Do we really need a yoke steering wheel, no shifter/wiper/turn signal stalks, or no radar? No. I'm all-in for innovation, but I feel they are just trying to be different for the sake of being different now. The Mach-E is really an amazing V1 EV, and I'm sure Ford will sell a ton of F150 Lightnings. I was thinking that VW would be the next dominant EV maker outside of China, but now I'm not sure...

For now, I will be absolutely holding off on any future software updates until this is figured out. As news like this spreads, I don't think I'll be alone. Tesla can't force it's customer's to downgrade their driver assist technologies, and this almost certainly will result in a class lawsuit.
I'm with you on delaying at least until this is sorted. It has really been poorly handled and it is unfortunate.

I'm much more optimistic on it long term though. The current system can't use radar to find lane lines and yet it works amazingly well even with poor markings at night in significant rain. If it can manage that, nothing should be stopping them from solving the rest in rain.
 
I'm with you on delaying at least until this is sorted. It has really been poorly handled and it is unfortunate.

I'm much more optimistic on it long term though. The current system can't use radar to find lane lines and yet it works amazingly well even with poor markings at night in significant rain. If it can manage that, nothing should be stopping them from solving the rest in rain.
Agreed - mine is on hold until August, but I might try to extend it further. Between this radar removal & the Jan 2022 federal EV credit just over the horizon right now, and potential for 4860 cells next year sometime, I'm leaning toward no reason to rush taking delivery this year, unless something drastically changes in the next few months. It'll suck to lose my low price from March reservation but I'd rather a car that's not freaking out at mist and road signs. Hopefully they don't roll out a yoke or remove more control stalks in 2022.

As excited as I was to get an EV, the Telluride, Pilot or Rav 4 Prime for the next 1-2 years doesn't seem horrible right now - radar, real wiper and shifter stalks, and give the Neural Net, EV market and congress more time to firm up.
 
  • Like
Reactions: dyau
Maybe this would be a good time to remind folks that cruise control and lane keeping systems routinely disengage for heavy rain and water on the road. At least if there is some form of active traction control in operation. The Cadillac XTS that I turned in for my Model 3 did that. My wife's Buick does the same. For that matter, both my Model 3 and the Model Y that I now own do the same (with radar). Both are FSD cars. Neither would allow me to use AP, or cruise, once bad weather was detected. The camera only system may be worse in this regard, but no cruise control systems that I have recent experience with will operate normally in compromised traction and visibility. In recent years Subaru has also relied on cameras. Road tests of these cars have commented favorably on their good-weather execution while some have marked them down in the rain. This is my memory talking BTW. My vote is to cut Tesla some slack here while we wait for full implementation.
 
I am very concerned about the automatic emergency braking in bad conditions. The adaptive cruise worked fine on my Acura even in rain. If you did put it in snow mode (of the dynamic driving modes it had) it would disable the cruise.

I remember one torrential downpour on a highway. No place to safely pull off. I just put the adaptive cruise control on max distance and kept going at a reasonable pace (not so fast the emergency braking or adaptive cruise control would hit someone nor so slow that someone would rear end me on the highway.
 
  • Like
Reactions: NikolaTs