Self driving update: Gen 2 and Gen 3 Autonomy on R2 should be similar for a "couple of years"

[Riviot]

I think your analogy has some room for improvement.

I'm looking at it like this:

You've met someone with a dating profile that says they want to have kids. The first date, they say "we're going to make beautiful kids together some day". You think that's a bit forward, but that's what you're looking for in a relationship, so you keep dating them. You date for a while, then try having kids. The first kid takes a few years of trying, comes out, and it's the ugliest kid you've ever seen. So you try again.

The second kid's even uglier than the first. You find out it's because of a genetic condition, that has a 100% chance of being transmitted to your children together, so you're never going to have good looking kids with your spouse. You start thinking about splitting up. Your spouse is suddenly outside of their warranty period, and starts having all kind of weird issues you decide you don't want to deal with anymore.

But then, you finally meet your spouse's older sibling, who already has a kid, and the kid is pretty goofy looking, but not nearly as ugly as your own kids.

You find out they are single, so that piques your interest. You get to know them a bit better, and find out they still want to have more kids. You decide to go all in, sell your current kids, and start dating the older sibling. Maybe the genetic condition won't be a concern this time, but it's too hard to know up front. Now, you've been dating for a while, and you're thinking about trying to have kids again. <---we are here

... You lost me at "kids"

... And got me back at selling kids.

Also, don't blame Rivian for your ugly kids. Or your divorces

Sponsored

 

[captainjp]

Looks like @pamalabama is back…
Moderators, start your engines.

 

[Dark-Fx]

... You lost me at "kids"

... And got me back at selling kids.

Also, don't blame Rivian for your ugly kids. Or your divorces

Guess you don't know what an analogy is.

 

[Riviot]

Guess you don't know what an analogy is.

 

[Zathras]

These comments have been overall consistent from Rivian Staff. Even at the Autonomy event, it was the overall conclusion I gathered from what was said.

However lets be clear, the Lidar & Gen 3 chip will be more future proof over time yet likely by the time it becomes relevant there's a high chance there will be a Gen 4 chip with a better lidar that will be presented as more future proof.

To me the Launch Package is a Great Value. I was on the fence but when he said Coastal Cloud will be part of the Launch Package I was sold.... Now let's just see how far away I am in this reservation line ‍ lol

Another factor will be if one plans to keep the car for more than two or three years. If not, such as if they are leasing it, no reason to wait. Same for those not enamored of level 4 driving anyway. Or at least the edge cases where LIDAR will eventually make somewhat of a difference. I only have about 20 or 30 years left of this life. And while I plan to do a lot of adventuring and camping in it, I'm hoping this could be my last car, and I want it to be able to take me to the doctor even if I'm blind. So I'm waiting.

Debbie Downer approves this message.

 

[Zathras]

This is the thing that everyone seems to glaze over. LIDAR may just be a mule for training but they’re adding a chip with 4x the performance of the current chip, 200 TOPS vs 800 TOPS. And then of course there’s two of each for redundancy. If we learn anything from Tesla it’s the compute power that is the issue as time moves along.

They have made it pretty clear that it's not just a mule for training. It will eventually be able to do more that cars not equipped with the gen 3 chip. But you're going to have to wait for several years for that to pay off. At first, it's going to be the mule. And I for one don't mind contributing to the flywheel. I hope eventually the government might create some standards and require all car companies to contribute to a massive single flywheel of truth that everyone benefits from. But I'm not holding my breath for several years to see that happen.

 

[shandering]

Another factor will be if one plans to keep the car for more than two or three years. If not, such as if they are leasing it, no reason to wait. Same for those not enamored of level 4 driving anyway. Or at least the edge cases where LIDAR will eventually make somewhat of a difference. I only have about 20 or 30 years left of this life. And while I plan to do a lot of adventuring and camping in it, I'm hoping this could be my last car, and I want it to be able to take me to the doctor even if I'm blind. So I'm waiting.

Debbie Downer approves this message.

There's basically no chance that a car you buy today from rivian will be your last self driving car

At minimum it seems like the cameras would need cleaning, and presumably the placement of the cameras on parts of the car that move like the side mirrors is a problem for both cleaning and the fact that they move

my belief is that no company will just have self driving out of the blue. You'll see it coming from a long way away and then they will build the car with the necessary hardware and specs to self drive

nobody will unlock millions of cars to overnight self drive. The rollout will be at the speed that it is easier to just manufacture new cars that can self-drive

 

[Mos Eisley]

There's basically no chance that a car you buy today from rivian will be your last self driving car

At minimum it seems like the cameras would need cleaning, and presumably the placement of the cameras on parts of the car that move like the side mirrors is a problem for both cleaning and the fact that they move

my belief is that no company will just have self driving out of the blue. You'll see it coming from a long way away and then they will build the car with the necessary hardware and specs to self drive

nobody will unlock millions of cars to overnight self drive. The rollout will be at the speed that it is easier to just manufacture new cars that can self-drive

Well... there's always a chance

" Global Average: Based on a global mortality rate of roughly 7.7–10 deaths per 1,000 people per year, the daily chance of you dying is about 0.002%. "

 

[captainjp]

Well... there's always a chance

" Global Average: Based on a global mortality rate of roughly 7.7–10 deaths per 1,000 people per year, the daily chance of you dying is about 0.002%. "

My take is that the current iteration of vehicles exist to train the systems that will ultimately provide fully autonomous vehicles. So, I agree with the previous comment to yours, however it is pretty obvious it will not happen overnight.

 

[Zathras]

There's basically no chance that a car you buy today from rivian will be your last self driving car

At minimum it seems like the cameras would need cleaning, and presumably the placement of the cameras on parts of the car that move like the side mirrors is a problem for both cleaning and the fact that they move

my belief is that no company will just have self driving out of the blue. You'll see it coming from a long way away and then they will build the car with the necessary hardware and specs to self drive

nobody will unlock millions of cars to overnight self drive. The rollout will be at the speed that it is easier to just manufacture new cars that can self-drive

Depends on how old you are, no?

Your assumptions about what is happening are quite off base. Out of the blue? The current vehicles will self-drive with time (a year or two, not many years into the future). We don't need all cars on planet Earth to be self-driving for it to be useful.

No doubt things will improve with future upgrades, but self-driving is already happening, and proven potentially safer than humans behind the wheel already. I imagine that will only improve if companies don't over-trust large language models and keep the hardware and software task-specific. Seems to me Rivian's a good way down the road on that already.

 

[shandering]

Depends on how old you are, no?

Your assumptions about what is happening are quite off base. Out of the blue? The current vehicles will self-drive with time (a year or two, not many years into the future). We don't need all cars on planet Earth to be self-driving for it to be useful.

No doubt things will improve with future upgrades, but self-driving is already happening, and proven potentially safer than humans behind the wheel already. I imagine that will only improve if companies don't over-trust large language models and keep the hardware and software task-specific. Seems to me Rivian's a good way down the road on that already.

The idea will be that let's say Tesla has a tech that they can roll out self driving to 500K cars over the period of a year. There's a lot that has to happen for this to work.

They can manufacture 500K cars a year. At that point it's easier and they are financially incentivized to just manufacture 500K new cars with the proper self driving hardware.

If all that is needed is camera cleaning that can be retrofit. tesla is retrofitting launch edition model Ys with camera cleaning hardware. When they were testing "robobtaxi" using cars that did not have camera cleaning it looks like they are retrofitting those

 

[Zathras]

Not sure where you are getting your information, but it's nowhere connected to reality. Telsa has the self-driving in their cars now, and after all these years it's working pretty good. Rivian has laid out their path to self-driving, and you simply have to look to see it. All this speculation about numbers of cars, and cleaners on cameras is utter gaslighting.

 

[shandering]

Not sure where you are getting your information, but it's nowhere connected to reality. Telsa has the self-driving in their cars now, and after all these years it's working pretty good. Rivian has laid out their path to self-driving, and you simply have to look to see it. All this speculation about numbers of cars, and cleaners on cameras is utter gaslighting.

The only cars running unsupervised in a tesla have camera cleaning. Tesla has billions of miles of FSD and they know more than anyone the bare minimum that is needed to do unsupervised

The other problem is when unsupervised is here there will be likely not many car makers left. I think the first person to have unsupervised with the right featureset will have a monopoly. When it costs 10 billion dollars a year to maintain the support to allow a consumer to drive themselves. you need a lot of cars to get the price affordable enough. Other car makers are done for.

https://www.fsddb.com/trackers/canadac2c

Currently there is a cross country road trip on FSD in canada that is happening and they are doing it 100% on FSD. They have had zero touching of the steering wheel or brakes, zero changing of driving profiles, only just entering the navigation

In order to do that it's not easy. They will supercharge 25+ times, the car needs to park perfectly at every single one. All of the superchargers seem to have detailed 3d maps with all stalls and which ones are out of order/being used. They are driving through rural/cities roads, construction zones that they have to reroute, they passed 2 construction flaggers in a row that the car had to understand the correct time to go and hten drive on the wrong side of the road.

At one point they were driving on at least a 4 lane highway with zero lane markings. the car sounded like it was driving on a rumble strip.

This is an example where another brand would be years from implementing this even if the tecnology is there.

 

[Zathras]

The only cars running unsupervised in a tesla have camera cleaning. Tesla has billions of miles of FSD and they know more than anyone the bare minimum that is needed to do unsupervised

The other problem is when unsupervised is here there will be likely not many car makers left. I think the first person to have unsupervised with the right featureset will have a monopoly. When it costs 10 billion dollars a year to maintain the support to allow a consumer to drive themselves. you need a lot of cars to get the price affordable enough. Other car makers are done for.

https://www.fsddb.com/trackers/canadac2c

Currently there is a cross country road trip on FSD in canada that is happening and they are doing it 100% on FSD. They have had zero touching of the steering wheel or brakes, zero changing of driving profiles, only just entering the navigation

In order to do that it's not easy. They will supercharge 25+ times, the car needs to park perfectly at every single one. All of the superchargers seem to have detailed 3d maps with all stalls and which ones are out of order/being used. They are driving through rural/cities roads, construction zones that they have to reroute, they passed 2 construction flaggers in a row that the car had to understand the correct time to go and hten drive on the wrong side of the road.

At one point they were driving on at least a 4 lane highway with zero lane markings. the car sounded like it was driving on a rumble strip.

This is an example where another brand would be years from implementing this even if the tecnology is there.

You are overstating (at least) what is reqïred to implement such a system. Rivian is going to have it done in not that many years. Their tech is improving at a pretty good rate. You believe what you want, but spewing massive word salad tomes doesn't make what's happening not true. And overstating Tesla's capabilities makes me wonder if we're talking with Grok.

 

[shandering]

You are overstating (at least) what is reqïred to implement such a system. Rivian is going to have it done in not that many years. Their tech is improving at a pretty good rate. You believe what you want, but spewing massive word salad tomes doesn't make what's happening not true. And overstating Tesla's capabilities makes me wonder if we're talking with Grok.

In order to self drive (unsupervised) you need the least amount of failures possilble. The entire U.S. needs to be mapped as accurately as possible. It reduces edge case issues with poor mapping. You need cameras that never get dirty (self cleaning) otherwise your car stops moving (minimum risk maneuver) which is a safety issue. During the 3500 mile FSD only drive, they had one failure around the 3500 mile mark where it was raining and the car pulled off to the side of the road with the hazards on. They had to get out and wipe down the camera. Not sure the camera was really all that dirty and it was on local roads, not a highway

The other issue is if the cameras are on parts of the car that move or in places that get dirty easily (like rivian side mirrors) then that probelm is even worse. You can see that the tesla camera placement is not the best, but it is designed around where they get the least dirty. it is also on parts of the car that don't move, so when you open the doors/whatever you never have to worry about alignment moving the position of the cameras.

you would also need a regular certification process multiple times a year to check the sensors and make sure everything works.

And then everywhere the car drives it would have to have validated performance, and the car's ODD will have to be strictly controlled.

Compared to FSD right now, rivian is at least 2-3 years from something resembling FSD 12 which would put them at 5 years to reach what tesla has now

And currently no one (outside of tesla) has demonstrated that they are even remotely close to doing unsupervised self driving without the use of HD maps, tight restrictions and all the typical robotaxi hardware.

There is no evidence that adding lidar shortcuts any sort of problem or adds any level of safety if you are not doing the traditional AV 1.0 approach

Sponsored