Rivian Explains Why CarPlay Debate Will Become 'Completely Obsolete'

[Thehangmn]

Every time I talk with people who say that lidar isn't an issue and that lidar is needed can't seem to provide any substance when I bring up Waymo's failures that LiDAR is designed to protect against.

ok. we can go off on the tangent.

1.Lidar measures distance. Where things actually are. a camera doesn't measure depth, only estimates it based on 2d pixels via parallax and motion between frames. Vision only systems consistently fail based on misreading flat frames.

2. Liadar works regardless of light conditions. It is illuminating the scene itself, it's an active sensor. The Camera on the other hand is a passive sensor and is only as good as the ambient light its handed. when you exit a tunnel into the sun, night driving. Cameras aren't perfect, and it only takes split seconds to crash a car.

3. Cameras being block or glare, have no back up. if your front facing cameras are disabled by glare. What data does the car use?

4.multiple cameras may be redundant, but its redundancy with the same technology. a single modality system will have a single fail point.

5. adverse weather conditions hit cameras hard. rain, fog, snow, dust... all disruptive to the camera sensors. lidar punches through. if you can't see... can you drive? when visibility collapses, lidar can carry on.


The claim that "humans drive with only vision" may be true, but our eyes are far more capable than the cameras, and they are connected to true general intelligence that can infer not only where things are, but the intent of other drivers. Even with all of that Humans crash cars quite often. so that bar is low. the Autonomous vehicle needs to be more capable than a human driver. we should not accept any less than perfect. so my question to you is why does vision alone seem adequate for you? there are clearly challenges that you have to address as well.

Your claim that contention is a problem is not incorrect, let's assume we have an autonomous vehicle with cameras and Lidar and that contention problem is solved, as problems can be... now what? as for Waymo they will fix the issue and the cars will be back on the freeway with an even more robust system.

now can we get back to CarPlay? why is CarPlay so important to you?

 

[Thehangmn]

1. Read the second part of the headline in your link.
2. Globally had a 6.3% increase in deliveries compared to Q1 2025.
3. Globally Q2 had a 25% increase in deliveries compared to Q2 2025.

It already bottomed.

Time will tell if what you say is real or just confirmation bias.

 

[germanbeer007]

ok. we can go off on the tangent.

1.Lidar measures distance. Where things actually are. a camera doesn't measure depth, only estimates it based on 2d pixels via parallax and motion between frames. Vision only systems consistently fail based on misreading flat frames.

2. Liadar works regardless of light conditions. It is illuminating the scene itself, it's an active sensor. T

I've developed apps using iPad's lidar. I've worked with the point clouds. I'm well versed in how lidar works.

The Camera on the other hand is a passive sensor and is only as good as the ambient light its handed. when you exit a tunnel into the sun, night driving. Cameras aren't perfect, and it only takes split seconds to crash a car.

3. Cameras being block or glare, have no back up. if your front facing cameras are disabled by glare. What data does the car use?

This is great, we can get down to the nitty gritty technical details.

Tesla's camera sensors have 19 stops of dynamic range (12 bit sensors). Human at any given point typically sees 14 stops. Of course human eyes can adapt to different brightness to get the full 20 stops but Tesla's camera sensors can capture 19 stops at an instant.

There are plenty of samples by a Tesla hacker named "greentheonly" who ran tests to see how cameras look at the sun and in pitch black (plenty of details are kept).

Now, the HW3 chip operates on 10-bit which discards some info (and that's why you see video clips of Tesla cars failing in glare). But AI4 chip operates on the full 12-bit. Can it still get glares? Sure, but it'll fail at a point where even a human can't see in that glare.

TL;DR, camera sensors see better than a human. As long as it sees better than a human, it's enough for autonomy. It doesn't have to be perfect.

4.multiple cameras may be redundant, but its redundancy with the same technology. a single modality system will have a single fail point.

A Waymo doesn't drive with only lidar. That's not how it works. Waymo sensor fuses all the data and then runs inference on that data. The supposed benefit here is higher accuracy when a camera picks up on an object.

If the camera sensors completely fail, Waymo will simply stall. It will not switch into some backup mode and drive with only lidar sensors like usual.

5. adverse weather conditions hit cameras hard. rain, fog, snow, dust... all disruptive to the camera sensors. lidar punches through. if you can't see... can you drive? when visibility collapses, lidar can carry on.

Lidar does not "punch through" rain. In fact, it's even worse for a multimodal system because you have noise contributing to lidar point clouds and cameras. Then you have sensor contention where lidar might make out an object that a camera system doesn't pick up on. Do you trust lidar or trust cameras?

Sensor contention is what caused Waymo to crash into a telephone pole. Lidar picked up a telephone pole but the camera system didn't really pick up on a super thin object so Waymo "assigned a low damage score" on the telephone pole and rammed into it.

The claim that "humans drive with only vision" may be true, but our eyes are far more capable than the cameras

Untrue. Explained above.

the Autonomous vehicle needs to be more capable than a human driver.

sure. an autonomous system is aware 24/7 with 360 degree vision.

we should not accept any less than perfect.

100% disagreed. millions of people die every year from car crashes. if a system reduces the death by 99.9%, you're still going to hold off on deploying a system because you need to chance that 0.1%?

That makes no sense. And Waymo, the thing you talk so highly of, crashes/stalls often.

so my question to you is why does vision alone seem adequate for you? there are clearly challenges that you have to address as well.

I've explained above.

Your claim that contention is a problem is not incorrect, let's assume we have an autonomous vehicle with cameras and Lidar and that contention problem is solved

If you can "assume" that, you can "assume" it's solved with cameras only.

, as problems can be... now what? as for Waymo they will fix the issue and the cars will be back on the freeway with an even more robust system.

And Tesla can solve issues and deploy on freeways too.

now can we get back to CarPlay? why is CarPlay so important to you?

You really didn't address the Waymo hitting a telephone pole. You just made an argument saying "assume that content problem is solved". That's not a proper response to the issue I'm pointing out.

 

[germanbeer007]

Time will tell if what you say is real or just confirmation bias.

What I gave were real delivery numbers straight from the delivery reports.

 

[I7guy]

ok. we can go off on the tangent.

1.Lidar measures distance. Where things actually are. a camera doesn't measure depth, only estimates it based on 2d pixels via parallax and motion between frames. Vision only systems consistently fail based on misreading flat frames.

2. Liadar works regardless of light conditions. It is illuminating the scene itself, it's an active sensor. The Camera on the other hand is a passive sensor and is only as good as the ambient light its handed. when you exit a tunnel into the sun, night driving. Cameras aren't perfect, and it only takes split seconds to crash a car.

3. Cameras being block or glare, have no back up. if your front facing cameras are disabled by glare. What data does the car use?

4.multiple cameras may be redundant, but its redundancy with the same technology. a single modality system will have a single fail point.

5. adverse weather conditions hit cameras hard. rain, fog, snow, dust... all disruptive to the camera sensors. lidar punches through. if you can't see... can you drive? when visibility collapses, lidar can carry on.


The claim that "humans drive with only vision" may be true, but our eyes are far more capable than the cameras, and they are connected to true general intelligence that can infer not only where things are, but the intent of other drivers. Even with all of that Humans crash cars quite often. so that bar is low. the Autonomous vehicle needs to be more capable than a human driver. we should not accept any less than perfect. so my question to you is why does vision alone seem adequate for you? there are clearly challenges that you have to address as well.

Your claim that contention is a problem is not incorrect, let's assume we have an autonomous vehicle with cameras and Lidar and that contention problem is solved, as problems can be... now what? as for Waymo they will fix the issue and the cars will be back on the freeway with an even more robust system.

now can we get back to CarPlay? why is CarPlay so important to you?

These are easy to find on the internet:
1. In Austin Waymo drove into flooded waters with a passenger.
2. Waymo has run red lights
3. In SF the entire Waymo fleet was incapacitated
4. Waymo drives wrong way down one way street

Anyway stories abound, no tech is perfect but the proof is in the pudding. FSD scales better than Waymo’s system.

Now, back to slugging it out about CarPlay.

 

[Thehangmn]

This is great, we can get down to the nitty gritty technical details.

Tesla's camera sensors have 19 stops of dynamic range (12 bit sensors). Human at any given point typically sees 14 stops. Of course human eyes can adapt to different brightness to get the full 20 stops but Tesla's camera sensors can capture 19 stops at an instant.

They can still suffer from blooming (blurring from extreme lights) and washed-out shadows when transitioning through environments like tunnels or facing direct sunlight.

There are plenty of samples by a Tesla hacker named "greentheonly" who ran tests to see how cameras look at the sun and in pitch black (plenty of details are kept).

Now, the HW3 chip operates on 10-bit which discards some info (and that's why you see video clips of Tesla cars failing in glare). But AI4 chip operates on the full 12-bit. Can it still get glares? Sure, but it'll fail at a point where even a human can't see in that glare.

TL;DR, camera sensors see better than a human. As long as it sees better than a human, it's enough for autonomy. It doesn't have to be perfect.



A Waymo doesn't drive with only lidar. That's not how it works. Waymo sensor fuses all the data and then runs inference on that data. The supposed benefit here is higher accuracy when a camera picks up on an object.

If the camera sensors completely fail, Waymo will simply stall. It will not switch into some backup mode and drive with only lidar sensors like usual.

I didn't say it would default to lidar only. but the data is constructed from redundancy sources. This is different from your claim on redundancy on tesla.

Lidar does not "punch through" rain. In fact, it's even worse for a multimodal system because you have noise contributing to lidar point clouds and cameras. Then you have sensor contention where lidar might make out an object that a camera system doesn't pick up on. Do you trust lidar or trust cameras?

sorry what is the camera only alternative? trust that the camera that can't see anything is actually seeing something?

Sensor contention is what caused Waymo to crash into a telephone pole. Lidar picked up a telephone pole but the camera system didn't really pick up on a super thin object so Waymo "assigned a low damage score" on the telephone pole and rammed into it.

So the Camera didn't pick it up.... so it would've rammed into it anyway?

Untrue. Explained above.



sure. an autonomous system is aware 24/7 with 360 degree vision.


100% disagreed. millions of people die every year from car crashes. if a system reduces the death by 99.9%, you're still going to hold off on deploying a system because you need to chance that 0.1%?

safer is always better... but until its perfect the goal should always be better.

That makes no sense. And Waymo, the thing you talk so highly of, crashes/stalls often.

So does the Robotaxi, arguably more often.

I've explained above.



If you can "assume" that, you can "assume" it's solved with cameras only.

except Teslas crash all the time and the robotaxi is a mess. You can say it's "solved" but there are teslas hitting poles too.

And Tesla can solve issues and deploy on freeways too.

Waymo CAN go on freeways. They chose to keep them from doing so for the sake of safety. Did tesla ever chose safety or did they let their customers be beta testers and crash?

You really didn't address the Waymo hitting a telephone pole. You just made an argument saying "assume that content problem is solved". That's not a proper response to the issue I'm pointing out.

Tesla Cybertruck crash on Full Self-Driving v13 goes viral

A Tesla owner reported that he crashed his Cybertruck into a pole after hitting a curb while using Full Self-Driving,...

electrek.co

teslas hit telephone poles too apparently? looks like camera only can't work.

 

[Thehangmn]

These are easy to find on the internet:
1. In Austin Waymo drove into flooded waters with a passenger.
2. Waymo has run red lights
3. In SF the entire Waymo fleet was incapacitated
4. Waymo drives wrong way down one way street

Anyway stories abound, no tech is perfect but the proof is in the pudding. FSD scales better than Waymo’s system.

Now, back to slugging it out about CarPlay.

do a google search about robotaxi? Tesla hasn't solved anything

 

[I7guy]

do a google search about robotaxi? Tesla hasn't solved anything

So because in your opinion Tesla hasn’t solved anything it’s okay for Waymo to drive into flooded waters and run red lights and have an incapacitated fleet? None of this is cogent.

 

[Thehangmn]

So because in your opinion Tesla hasn’t solved anything it’s okay for Waymo to drive into flooded waters and run red lights and have an incapacitated fleet? None of this is cogent.

yep, you got me all figured out.

Waymo has problems to solve, so does Tesla. Y'all present this thing like Tesla has it all figured out and solved. They don't. Their Robotaxi fleet has gobs of issues. you can keep pulling out individual issues, but a similar tesla issues is a google search away. Vision only hasn't solved the autonomous driving problem, and adding lidar doesn't solve it either. both approaches will continue to move forward and both are already claiming better safety than human drivers.

as I recall my original point had something to do with waymo not being an auto maker so i suggested the next step for them is to make their autonomous driving suite available to OEMs, and tesla is unlikely to do this.

 

[diamond.g]

yep, you got me all figured out.

Waymo has problems to solve, so does Tesla. Y'all present this thing like Tesla has it all figured out and solved. They don't. Their Robotaxi fleet has gobs of issues. you can keep pulling out individual issues, but a similar tesla issues is a google search away. Vision only hasn't solved the autonomous driving problem, and adding lidar doesn't solve it either. both approaches will continue to move forward and both are already claiming better safety than human drivers.

as I recall my original point had something to do with waymo not being an auto maker so i suggested the next step for them is to make their autonomous driving suite available to OEMs, and tesla is unlikely to do this.

Tesla has claimed FSD is available for use by other OEM’s. There was rumor of a deal being made, but it allegedly fell through because the interested party wanted changes that Tesla wasn’t willing to make.

 

[I7guy]

yep, you got me all figured out.

Waymo has problems to solve, so does Tesla. Y'all present this thing like Tesla has it all figured out and solved. They don't. Their Robotaxi fleet has gobs of issues. you can keep pulling out individual issues, but a similar tesla issues is a google search away. Vision only hasn't solved the autonomous driving problem, and adding lidar doesn't solve it either. both approaches will continue to move forward and both are already claiming better safety than human drivers.

as I recall my original point had something to do with waymo not being an auto maker so i suggested the next step for them is to make their autonomous driving suite available to OEMs, and tesla is unlikely to do this.

Y’all presented this like Waymo has it all figured out and Waymo’s system is “better” than Tesla’s. It isn’t. Pulling out individual issues while presenting straw man arguments doesn’t prove a point because each has its share of problems as manufacturers race to solve issues within their domain.

 

[Thehangmn]

You really didn't address the Waymo hitting a telephone pole. You just made an argument saying "assume that content problem is solved". That's not a proper response to the issue I'm pointing out.

quick revisit on this. Waymo's own NHTSA recall (campaign 24E‑049) attributes the May 2024 Phoenix pole collision to a software‑plus‑mapping defect. The system assigned the pole a low "damage score" and the map hadn't encoded a hard road edge in that curbless alley, so the car had insufficient ability to avoid a permanent object sitting within the drivable surface. your earlier characterization that waymo "rammed" into the pole is a little over stated given that it was going 8mph. if you want to see what Ramming looks like see the article about the cybertruck hitting the pole above

I can't find any source, primary or secondary, that attributes the crash to a lidar‑versus‑camera sensor disagreement. So the "lidar saw the pole, the camera missed it, and contention caused the crash" story doesn't hold water. The documented cause is planning/cost‑modeling and maps.
Sensor contention is a very real and difficult problem, but this crash isn't a clean example of it.

 

[Thehangmn]

Tesla has claimed FSD is available for use by other OEM’s. There was rumor of a deal being made, but it allegedly fell through because the interested party wanted changes that Tesla wasn’t willing to make.

Tesla claims a lot of things, lets assume this is true, my earlier point entertained this, and my conclusion was that competitors would have issues with being reliant on a direct competitor. Some have claimed this is one of the many reasons that apple killed its car project. if apple put out a car you can bet that OEMs would end CarPlay support immediately.

 

[Thehangmn]

Y’all presented this like Waymo has it all figured out and Waymo’s system is “better” than Tesla’s. It isn’t. Pulling out individual issues while presenting straw man arguments doesn’t prove a point because each has its share of problems as manufacturers race to solve issues within their domain.

Fair, I presented it like waymo is considered to be the leader in the space, and broadly considered to be better. I maintain this is true, but obviously Tesla fans will disagree.

 

[germanbeer007]

They can still suffer from blooming (blurring from extreme lights) and washed-out shadows when transitioning through environments like tunnels or facing direct sunlight.

Cameras transition when they need to adjust ISO/shutter to change the dynamic range. When you have 19 stops of dynamic range, this is generally not needed (especially leaving tunnels or most situations facing bright light). But if it does, the cameras on Tesla can "transition" nearly instantly because there's nothing mechanical about changing iso/shutter. There is no aperture to change.

I didn't say it would default to lidar only. but the data is constructed from redundancy sources. This is different from your claim on redundancy on tesla.

Your assumption is "if camera system fails" so the only other sensors are Lidar and radar. Radar doesn't contribute much to the reconstruction of the scene.

sorry what is the camera only alternative? trust that the camera that can't see anything is actually seeing something?

You seem to be under the impression that Waymo drives in the rain perfectly.

Waymo and Tesla both pull over to the side if visibility is too obstructed. DMV handbook tells human drivers to pull over if visibility is terrible. Even Waymo had to shutdown in Florida due to the rain while Tesla Robotaxi literally launched their service in Florida that same rainy day.

So the Camera didn't pick it up.... so it would've rammed into it anyway?

What? Camera sensors obviously can see the telephone pole. But the inference from lidar/camera fusion failed to score the pole as a high collision damage score so Waymo rammed into it. The reason why they're using "damage scores" is to ignore all the false positives that lidar picks up and set a safe threshold for Waymo to ignore (effectively filtering out lidar noise). Tesla doesn't do this at all.

safer is always better... but until its perfect the goal should always be better.

you didn't answer the question. are you saying Waymo/Tesla should chase that final 0.1% of safety? You do know that it's impossible to reach 100%, statistically speaking, right?

So does the Robotaxi, arguably more often.

1. But you seem to accept Waymo being on roads and you just said "we should not accept any less than perfect". You contradicted yourself
2. More often is objectively false considering Waymo has more cars.

except Teslas crash all the time and the robotaxi is a mess. You can say it's "solved" but there are teslas hitting poles too.

Waymo is doing all of those failures on a larger scale autonomously than Tesla. Yet you seem to be defending Waymo. You're contradicting yourself.

Waymo CAN go on freeways. They chose to keep them from doing so for the sake of safety. Did tesla ever chose safety or did they let their customers be beta testers and crash?

Tesla has never ran robotaxi driverlessly for customers, ever for freeways.
Waymo ran freeways driverlessly for customers and then pulled the service after they realized it didn't work. https://abc7news.com/post/waymo-rec...eway-work-zones-due-software-glitch/19325191/. This literally contradicts your insinuation Waymo chose safety.

You seem to be trying to talk about ADAS FSD Supervised which is irrelevant to the discussion.

Tesla Cybertruck crash on Full Self-Driving v13 goes viral

A Tesla owner reported that he crashed his Cybertruck into a pole after hitting a curb while using Full Self-Driving,...

electrek.co

teslas hit telephone poles too apparently? looks like camera only can't work.

1. Again, "You really didn't address the Waymo hitting a telephone pole.". You're pulling whataboutism.
2. We've seen many reports of teslas crashing and drivers blaming FSD when it turns out the evidence showed they were lying. One person in Australia even went to jail for lying about it. https://www.abc.net.au/news/2024-05-10/sakshi-agrawal-tesla-autopilot-crash-sentencing/103829772

Please address Waymo hitting a telephone pole. Do not pull whataboutism.

 

[germanbeer007]

quick revisit on this. Waymo's own NHTSA recall (campaign 24E‑049) attributes the May 2024 Phoenix pole collision to a software‑plus‑mapping defect. The system assigned the pole a low "damage score" and the map hadn't encoded a hard road edge in that curbless alley, so the car had insufficient ability to avoid a permanent object sitting within the drivable surface. your earlier characterization that waymo "rammed" into the pole is a little over stated given that it was going 8mph. if you want to see what Ramming looks like see the article about the cybertruck hitting the pole above

Overstated? I literally said it was going 8MPH https://forums.macrumors.com/thread...e.2482999/page-21?post=34667001#post-34667001 .

And I did mention the low damage score. https://forums.macrumors.com/thread...ome-completely-obsolete.2482999/post-34675374

You're stating things I already told you.

Even at that low speed, the damage sure does look like the car rammed into the pole. Look at the damage.

And you're still defending Waymo despite this clear evidence it's not up to your 100% safe threshold.

I can't find any source, primary or secondary, that attributes the crash to a lidar‑versus‑camera sensor disagreement. So the "lidar saw the pole, the camera missed it, and contention caused the crash" story doesn't hold water.

It does hold water if you understand the Waymo system as explained from the papers they put out.

We know a lidar system can detect in broad daylight the distance of objects within 1-3 cm of accuracy. The camera system (this includes inference) can also pickup objects but not with certainty. Had Waymo listened to the lidar data alone, it would have avoided it. But because it was fused with cameras and radar, a low damage score is evidence that not lidar but camera or radar contributed to a lower damage score. That's sensor contention. One system's data contributes to high, but other systems contribute lower damage scores.

The documented cause is planning/cost‑modeling and maps.
Sensor contention is a very real and difficult problem, but this crash isn't a clean example of it.

No. The fundamental issue was the low damage score which deals with sensor contention. The mapping update is basically writing a special case on maps to particularly avoid a pole at certain areas for extra caution, it doesn't change the actual fundamental issue

 

[Thehangmn]

What I gave were real delivery numbers straight from the delivery reports.

as I said, you gave numbers from a single quarter.. the rest is conjecture.

Overstated? I literally said it was going 8MPH https://forums.macrumors.com/thread...e.2482999/page-21?post=34667001#post-34667001 .

And I did mention the low damage score. https://forums.macrumors.com/thread...ome-completely-obsolete.2482999/post-34675374

You're stating things I already told you.

Even at that low speed, the damage sure does look like the car rammed into the pole. Look at the damage.

And you're still defending Waymo despite this clear evidence it's not up to your 100% safe threshold.



It does hold water if you understand the Waymo system as explained from the papers they put out.

We know a lidar system can detect in broad daylight the distance of objects within 1-3 cm of accuracy. The camera system (this includes inference) can also pickup objects but not with certainty. Had Waymo listened to the lidar data alone, it would have avoided it. But because it was fused with cameras and radar, a low damage score is evidence that not lidar but camera or radar contributed to a lower damage score. That's sensor contention. One system's data contributes to high, but other systems contribute lower damage scores.



No. The fundamental issue was the low damage score which deals with sensor contention. The mapping update is basically writing a special case on maps to particularly avoid a pole at certain areas for extra caution, it doesn't change the actual fundamental issue

youre getting worked up. I didn’t say that anything less than perfect shouldn’t be on the road, it’s that we should always strive to close the gaps in safety. If lidar gives data that cameras don’t then it improves the dataset. Your point seems to be that more data is always bad because it means you have to “decide” what to do when the sensors don’t align. But you ignore the obvious point that LiDAR can be right when cameras are not. When cameras are wrong and that’s all the data you have… the system will always make the wrong choice.

I don’t deny that LiDAR and cameras will have contention, but that is ultimately a problem that is partially solved. Contention doesn’t always cause issues.

You said the camera didn’t see the pole and the lidar did. It’s clear that if the lidar saw it the system should’ve trusted lidar more. I don’t see anywhere in the sources that describe the contention as the cause of the low damage score, and certainly no blame attributed specifically to LiDAR. Now can you explain how vision only what’s so good that the cybertruck crashed into a pole going much faster? In your example you call out lidar as the problem. So in the Tesla example you will certainly allow me to call out the lack of lidar as the cause, given that the cybertruck definitely was in FSD and hit the pole.

 

[germanbeer007]

as I said, you gave numbers from a single quarter.. the rest is conjecture.

Exact words were: "Time will tell if what you say is real or just confirmation bias.". The numbers are real.
And it wasn't just a single quarter.

youre getting worked up. I didn’t say that anything less than perfect shouldn’t be on the road,

You literally said "we should not accept any less than perfect". Exact quote.

I don’t deny that LiDAR and cameras will have contention, but that is ultimately a problem that is partially solved. Contention doesn’t always cause issues.

Well if it's not solved yet, they shouldn't be on the road, under your requirement of "we should not accept any less than perfect".

You said the camera didn’t see the pole and the lidar did. It’s clear that if the lidar saw it the system should’ve trusted lidar more. I don’t see anywhere in the sources that describe the contention as the cause of the low damage score, and certainly no blame attributed specifically to LiDAR.

Waymo PR isn't going to spell it out for you about how exactly it failed because it's their job to minimize the downsides of their system publicly. But there's no other practical reason why the damage score was lower than it should be. Sensors piped the data to the chip handling inference. Either lidar inference didn't pick up on the telephone pole (unlikely) or the camera inference didn't pick up on the telephone pole (very likely).

Now can you explain how vision only what’s so good that the cybertruck crashed into a pole going much faster?

User error.

 

[diamond.g]

Tesla claims a lot of things, lets assume this is true, my earlier point entertained this, and my conclusion was that competitors would have issues with being reliant on a direct competitor. Some have claimed this is one of the many reasons that apple killed its car project. if apple put out a car you can bet that OEMs would end CarPlay support immediately.

I have no clue if it is true, they did mention it on an investor call, so I am not sure if they would be willing to flat out lie to investors about it.

I do wonder how much upkeep CP requires. Especially since this would have to be in addition to whatever in house system they have to maintain. Word on the street was Tesla hasn't added CP cause it would screw with FSD, which allegedly Apple has fixed by being able to tell the car where to go. I wonder if Rivian would change their minds (they too are working on a point to point system that is supposed to be out this year) once Tesla adds it.

 

[Charlie Bonesx]

Tesla has never ran robotaxi driverlessly for customers

nope

 

[Thehangmn]

Exact words were: "Time will tell if what you say is real or just confirmation bias.". The numbers are real.
And it wasn't just a single quarter.

I’m sure your confidence is real and not biased.

You literally said "we should not accept any less than perfect". Exact quote.

Which I stand by, we should always push for 100% safety. That is the goal of safety systems my dude. You realize that right?

Well if it's not solved yet, they shouldn't be on the road, under your requirement of "we should not accept any less than perfect".

You have a hard time with this one. The idea of not accepting anything less than perfect means we should always be pushing for better safety. Saying “this is good enough, we don’t need redundant systems”. Isn’t safety mindset.

Waymo PR isn't going to spell it out for you about how exactly it failed because it's their job to minimize the downsides of their system publicly. But there's no other practical reason why the damage score was lower than it should be. Sensors piped the data to the chip handling inference. Either lidar inference didn't pick up on the telephone pole (unlikely) or the camera inference didn't pick up on the telephone pole (very likely).

Conjecture.

User error.

Or… it was a FSD error. But you couldn’t ever admit that.

 

[germanbeer007]

I’m sure your confidence is real and not biased.

You were talking about what I said. I said nothing about "confidence".
Seems like you're moving the goal post to something.

Which I stand by, we should always push for 100% safety. That is the goal of safety systems my dude. You realize that right?

You said we shouldn't "accept any less than perfect". Saying push for 100% safety changes your earlier meaning, my dude.

You have a hard time with this one. The idea of not accepting anything less than perfect means we should always be pushing for better safety. Saying “this is good enough, we don’t need redundant systems”. Isn’t safety mindset.

You can't say you accept a Waymo being on the road today while at the same time say "we shouldn't accept any less than perfect". They're contradictory.

At the same time I can absolutely say "I accept Tesla Robotaxi currently on the road but also push for 100% safety". They don't contradict each other. See where you went wrong?

Conjecture.

No. Simply applied logic.

Or… it was a FSD error. But you couldn’t ever admit that.

Or... it was user error. But you couldn't ever admit that.

 

[Charlie Bonesx]

Or... it was user error. But you couldn't ever admit that.

Or… it was FSD error. But you could never admit that.

 

[I7guy]

Or… it was FSD error. But you could never admit that.

The last crash was user error.

 

[Charlie Bonesx]

The last crash was user error.

If you trust Tesla…

Though regardless of whichever crash you are calling the last one, the example being discussed was FSD

they both have a long way to go but waymo is much further along