Lidar vs Camera revisited

[Bladerskb]

At this point, I'm fully convinced vision-only FSD will work and is the best approach to achieve generalized FSD first. In the future, we may have improved and cheap sensors, but for now, Tesla will win this race to a consumer-deployable generalized (90%+ of USA) robotaxi service.

I'm convinced because what Tesla is able to do with their approach and diverse data advantage. Here's one example:

https://twitter.com/x/status/1467198613298114561

Tesla is able to implement state-of-the-art approaches on the fly, test them out in their diverse fleet in all weather conditions and locales, and then re-deploy / fine-tune / unit-test on the fly. Tesla has the best iterative approach to a problem that no one has solved. Their iterations are 100x (my hyperbole) faster and more data-based than any competitor.

This is a complete myth borne by actual ignorance of what others are doing. The facts are pretty clear when you compare the content of AI DAY versus what others have presented. Tesla literally doesn’t use RL. While for others like Cruise and mobileye, RL is a huge part of their driving policy.

Tesla clearly doesn’t deal with uncertainty or reasoning which explains some of the dumb things it does while Cruise has that built into their NN architecture.

The difference in NN complexity and advancement is literally light years.

I could go on and on.
Tesla is actually behind others in their stack moving to software 2.0. They literally tell you that. It’s amazing how ignorance is bliss and people chose spreading balant lies.

 

[EVNow]

The difference in NN complexity and advancement is literally light years.

You don’t even know the meaning of “literally” and “light years” let alone insider info of state of the art of multiple companies.

 

[Bladerskb]

Thats because you haven't looked. Every AV company / legacy auto maker claimed FSD would be here by now. Infact they all said "in 5 years" - more than 5 years back.

Infact where Elon is consistently wrong is in predicting "1 year" instead of "5 years".

Except the big difference being others were predicting L4 in a geofenced city which actually happened in 2020. Meaning they were -1-2 years off based on which statement you are analyzing. Elon on the other hand was claiming Level 5 by Jan 2018. Let’s say Tesla magically gets to L5 in 2028. That would still make him off by over a decade.

You don’t even know the meaning of “literally” and “light years” let alone insider info of state of the art of multiple companies.

None of this is insider info. All of it is out in the open. The problem is that people REFUSE to be educated. I have talked to many Tesla fans who tweet by the min/hour daily about how amazing FSD is and NNs and no one else does this or that. How everyone should watch autonomy day and AI DAY. When you ask them have they watch ANYTHING about what any company tech consists of or what they are actually doing. They say no. Then when you suggest for them to watch a video, they refuse.

So you have a large group of people that are purposely ignorant.

 

[okcazure]

Except the big difference being others were predicting L4 in a geofenced city which actually happened in 2020. Meaning they were -1-2 years off based on which statement you are analyzing. Elon on the other hand was claiming Level 5 by Jan 2018. Let’s say Tesla magically gets to L5 in 2028. That would still make him off by over a decade.


None of this is insider info. All of it is out in the open. The problem is that people REFUSE to be educated. I have talked to many Tesla fans who tweet by the min/hour daily about how amazing FSD is and NNs and no one else does this or that. How everyone should watch autonomy day and AI DAY. When you ask them have they watch ANYTHING about what any company tech consists of or what they are actually doing. They say no. Then when you suggest for them to watch a video, they refuse.

So you have a large group of people that are purposely ignorant.

I'm interested in all L2/L3/L4/L5 technology, not just Tesla. Please link some videos on "Cruise and mobileye" that you are referring to. I regularly watch videos on Tesla, FSD, and others including comma – introducing comma three and Waymo. I'm more than happy to watch some additional videos! I have certainly seen mobileye videos in the past but never cruise, and it's been a while for mobileye. I'm sure you can link in some recent videos to showcase their current state. Thanks!

 

[Bladerskb]

I'm interested in all L2/L3/L4/L5 technology, not just Tesla. Please link some videos on "Cruise and mobileye" that you are referring to. I regularly watch videos on Tesla, FSD, and others including comma – introducing comma three and Waymo. I'm more than happy to watch some additional videos! I have certainly seen mobileye videos in the past but never cruise, and it's been a while for mobileye. I'm sure you can link in some recent videos to showcase their current state. Thanks!

Cruise...

 

[clydeiii]

How do I buy a Cruise car that can drive me? That's the bar. I can buy a Tesla w FSD right now. Sure, it might not work 100% perfectly, but it's surprisingly good and getting better all the time.

 

[EVNow]

How do I buy a Cruise car that can drive me? That's the bar. I can buy a Tesla w FSD right now. Sure, it might not work 100% perfectly, but it's surprisingly good and getting better all the time.

No point talking to concern trolls ...

 

[okcazure]

Cruise...

Thanks for sharing the link.. I'm watching now . It's always interesting to see the different approaches to the same problem.

 

[_Redshift_]

Thats good information. It also makes you an employee of potential Tesla competitor. You should note that in your signature.


Depends on your idea of what the car error rate can be. If you think about it - there is so much chaos on the road that there should be a lot more accidents. And then you go and see real chaos on the roads of Cairo or Chennai. Then you wonder how come there aren't accidents every second in every road.


Those in the "field" also thought you can't make profitable EVs or make people actually want to buy EVs. 'nuff said.

You are just saying the same thing we have all been talking about for years here. Nothing new.

ps : working on projects that you are sure will fail, is one way to make sure the project fails.

Um, there ARE a lot of accidents in Chennai. Or other cities.

I have spent over 20 years in Hardware design verification and know a lot about corner cases. There are way too many on the road. I have always been of the opinion that robust FSD is a pipe dream, at least for the next 8-10 years.

Proof? My 2018 Model 3 with EAP still has phantom braking, trouble deciding whether to exit the carpool lane or remain in it, even with the destination on the nav being more than 15 miles away, and many more.

 

[drtimhill]

I wonder if we will get lidar by hw5. Maybe it will cost $20 by then. @Bladerskb says it costs $40 each now?

Day X: Tesla adds Lidar to HW
Day X+1: Hundred of people complain about new phantom braking.

 

[Terminator857]

I didn’t say it was $40. I said nio ET5’s cost $40k.

High resolution lidar in mass production today cost anywhere between $150 to $500

Just to clarify. I was wrong. Cost is as stated above: $150+
Looking forward to when price is $50 in massive quantities. If we invoke Moores law, then price halves every couple of years, so maybe in 4 years. Yeah it is a stretch to apply Moores law to a laser, but electronics often see price drops also, often do to more mass manufacturing, using lower cost components, patent expiration and other tricks to lower price.

 

[S4WRXTTCS]

Day X: Tesla adds Lidar to HW
Day X+1: Hundred of people complain about new phantom braking.

To my knowledge Lidar doesn't introduce phantom braking events.

Radar has known phantom braking due to a noisy signal, and lack of resolution (especially what Tesla uses).

Vision has known phantom braking with things like shadows. It took Subaru years to evolve their eyesight into something that works reasonably well.

My biggest issue with Phantom braking for awhile now hasn't been due a sensor, but other sources. Sources such as Nav issues or the car suddenly tracking the car in the lane next to mine for who knows what reason.

Heck sometime I'll get phantom phantom braking. Meaning the car will flash something like "stopping for signal" on a freeway where it never actually phantom brakes. It just warns me that its highly considering it.

 

[EVNow]

Um, there ARE a lot of accidents in Chennai. Or other cities.

True - but if it was pure chaos as you say, there should be a lot more.

I have always been of the opinion that robust FSD is a pipe dream, at least for the next 8-10 years.

Depending on your definition of robust (i.e. what ODD and what error rate) - 2030 looks like a good idea to check back. I've been saying that for years ...

But that doesn't mean we can't have useable L2 level "FSD" in the next couple of years.

 

[_Redshift_]

True - but if it was pure chaos as you say, there should be a lot more.


Depending on your definition of robust (i.e. what ODD and what error rate) - 2030 looks like a good idea. I've been saying that for years ...

One accident can hurt you badly or kill you. What’s your standard?

Robust, to me is - no phantom braking. At all. None. Zero. I don’t do braking with emergency force for no reason, neither should the software. No ping-ponging between exit carpool lane or stay in the lane when the navigation knows the destination is very far away.

No yanking the steering hard to take the car away from a merging lane onto the freeway with no other car in the adjacent lane.

No wipers going randomly crazy on me. They have worked for 20+ years fine, until you Tesla engineers came along and screwed them up.

No suggesting a camera is blocked just because there is bright sunshine (no dirt) on the camera. Add a camera cleaning system, so the system doesn’t degrade in inclement weather.

For heavens sake, get out of the ‘cameras are enough’ nonsense. They don’t work well for simple wiper controls, jeez. To make that trope stick, Tesla is having to keep the auto high beams on at night. What nonsense!

Sorry, but I was actually pretty staunch fan of Tesla and Elon until recently. I still feel strongly about them, and would want them to succeed. However their culture after sampling success has become too arrogant and careless.

 

[Matias]

When will we know if he was incorrect?

Maybe if there are consumer level 3 cars on the market with lidar and Tesla is still level 2?

 

[EVNow]

One accident can hurt you badly or kill you. What’s your standard?

If you don’t want accidents at all - you need to not go out of the house. Oh - you can have accidents in the house too, oops.

We can have serious discussion when you get real

 

[EVNow]

Maybe if there are consumer level 3 cars on the market with lidar and Tesla is still level 2?

You mean consumer cars where L3 is not limited to below 37 mph on select 60 mph freeways

 

[Daniel in SD]

Maybe if there are consumer level 3 cars on the market with lidar and Tesla is still level 2?

Nope, only when another manufacturer releases a “L5 no geofence” vehicle that works on every road on earth and costs less than the lowest price Model 3 sold with FSD (no inflation adjustment allowed).
Also doesn’t count if it’s ICE.

 

[Matias]

You mean consumer cars where L3 is not limited to below 37 mph on select 60 mph freeways

Yes.

 

[marketisopen]

I don't understand how the more accurate point clouds of LIDAR could possibly be a disadvantage.

They are over specified so the computer is much slower because of that. So gathering needless information isn't good.