Rivian May Be Much Closer to FSD Than You Think

[Donald Stanfield]

I think the issue is people blame Tesla for beta testing on customers and deaths using FSD (find me them?) across 6.5 billion miles.

And somehow Rivian is taking the higher ground in delaying the release of half-assed software purely due to a skill issue.

When FSD was bad back in 2020, you had to have a safety score of 100 to get access. RIvian does not have any sort of safety score software at all. They will beta test on consumers. Mark my words.

Another delusional Tesla apologist spotted. Denying that FSD has killed anyone is pure delusion.

Sponsored

 

[sunydrm]

Do you have any data that supports this assertion?

Self driving costs billions of dollars to develop. Tesla spent more than 10 billion on training compute and storage. Rivian has historically been very bad at cheaply developing advanced techologies. Their spending has historically been very inefficient. Rivian does not have this money. This is purely to deliver L2++. The chinese are spending this kind of money.

Tesla is a tier 1 AI company and they have probably 800 people working on this problem. At 200K a year this adds up.

Delivering L4 self driving at scale (unsupervised) requires remote support centers all around the U.S. to have 24-7 low latency remote assistance.

You need perfect, lane accurate maps of the U.S. Without it you will not get required safety in poor visibility. The maps will also need to be real time updated with all road closures and events happening around the U.S., all school zone times, etc. The vehicles will need real time vehicle to vehicle communication

You will need 10x the number of service centers that Rivian has now. The cars will have to be serviced at least 4x a year to clean sensors, inspect road worthiness, and calibrate the lidar/radars. You can't just drive around unsupervised without doing this.

You need charging networks to enable road tripping, you need collision centers because these cars will not be accident-free

Then most importantly you will have to continue to spend yearly on developing self-driving or you will get left behind by other cars with more features

 

[sunydrm]

Another delusional Tesla apologist spotted. Denying that FSD has killed anyone is pure delusion.

Did I deny? I am just asking to see proof of these dozens of deaths

Tesla finally released more apples to apples safety data including a tracker of FSD usage. 6.5 billion miles with presumably 50% of accidents being not-at fault.

Data shows 2x improvement in serious accident rate. There will always be a limit where you cannot improve due to human drivers. In order to have a greater than 2x accident rate reduction you need to dodge accidents from the other party. Humans already do this and it is factored into the accident rate. You need superhuman skills.

It is not unreasonable to expect 30+ people to have died

If you look at robotaxi in austin there are 7 low speed accidents. As no surprise, 5 of them are not at fault. 2 of them are hitting some random thing in a parking lot at low speed

 

[mkhuffman]

Self driving costs billions of dollars to develop. Tesla spent more than 10 billion on training compute and storage. Rivian has historically been very bad at cheaply developing advanced techologies. Their spending has historically been very inefficient. Rivian does not have this money. This is purely to deliver L2++. The chinese are spending this kind of money.

Tesla is a tier 1 AI company and they have probably 800 people working on this problem. At 200K a year this adds up.

Delivering L4 self driving at scale (unsupervised) requires remote support centers all around the U.S. to have 24-7 low latency remote assistance.

You need perfect, lane accurate maps of the U.S. Without it you will not get required safety in poor visibility. The maps will also need to be real time updated with all road closures and events happening around the U.S., all school zone times, etc. The vehicles will need real time vehicle to vehicle communication

You will need 10x the number of service centers that Rivian has now. The cars will have to be serviced at least 4x a year to clean sensors, inspect road worthiness, and calibrate the lidar/radars. You can't just drive around unsupervised without doing this.

You need charging networks to enable road tripping, you need collision centers because these cars will not be accident-free

Then most importantly you will have to continue to spend yearly on developing self-driving or you will get left behind by other cars with more features

More assertions without real data.

I don't care, really. Your logic is reasonable, but still. It is all assertions without any facts to back them up.

 

[sunydrm]

More assertions without real data.

I don't care, really. Your logic is reasonable, but still. It is all assertions without any facts to back them up.

Look at tesla robotaxi service. Even to scale to 20 cities WITH safety monitors they have to build depots in every city they operate. Even the path to delivering 1000 cars on the robotaxi network is hard.

Elon mentioned in one earnings call that tesla will raise money to fund robotaxi if they need to. That is a company with (37? billion dollars) and they might need to fundraise money to scale self driving

The idea that these cars will overnight turn on and be self driving will never happen. And it's not a sensor issue.

 

[mkhuffman]

Look at tesla robotaxi service. Even to scale to 20 cities WITH safety monitors they have to build depots in every city they operate. Even the path to delivering 1000 cars on the robotaxi network is hard.

Elon mentioned in one earnings call that tesla will raise money to fund robotaxi if they need to. That is a company with (37? billion dollars) and they might need to fundraise money to scale self driving

The idea that these cars will overnight turn on and be self driving will never happen. And it's not a sensor issue.

I think that is a autonomous taxi issue, not a self driving issue. Rivian has never said they plan to offer robo-taxis.

 

[sunydrm]

I think that is a autonomous taxi issue, not a self driving issue. Rivian has never said they plan to offer robo-taxis.

Level 2++ is useless. What we are discussing in this thread is self driving aka UNSUPERVISED. Ignoring tesla's misleading use of the word FSD we are talking about unsupervised in this thread.

RJ says consumers won't buy a car without autonomy. He means unsupervised.

Delivering at scale instead of geofenced areas is even harder.

 

[captainjp]

Looks like @pamalabama has resurfaced under a new pseudonym…

 

[TM1]

GM is adding Lidar in 2028.

Then it's official, Lidar is dead. It's like Jim Cramer recommending a stock as a buy.

 

[TM1]

Another delusional Tesla apologist spotted. Denying that FSD has killed anyone is pure delusion.

It's also delusional to deny that FSD hasn't saved lives yet.

 

[nsuglerio]

Another delusional Tesla apologist spotted. Denying that FSD has killed anyone is pure delusion.

That system is only going to get better and if people assume responsibility and pay attention when using the system most of those accidents could be prevented. Tesla shifting the name from autopilot to "Full Self Driving" is the most damning indictment I think someone can make. Hence they started padding the (supervised).

In this very thread we have another guy who mentioned how he felt Rivian's system was dangerous. As I said previously, Rivian's Mobile-eye based solution (before their own rollout ) suddenly veered my R1T off an exit at highway speeds without any warning, but then I was still keeping my hands near the wheel and the brake press disabled the system in time.


The fact is these systems over time will save lives which people fail to mention:

National Safety Council:
"The study projects that the use of ADAS technologies could help avoid 249,400 fatalities and 14,138,000 nonfatal injuries from 2021 through 2050, in the most likely scenario. The most pessimistic scenario projects avoiding 152,100 fatalities and 8,673,000 nonfatal injuries. In comparison, the most optimistic scenario projects the avoidance of 298,300 fatalities and 16,814,000 nonfatal injuries. "

National Highway Traffic Safety Administration:
"NHTSA estimates that together these functions could potentially prevent up to nearly 600,000 crashes and about 1,300 fatalities annually when fully deployed"

This isn't even a Tesla apologist / vibes thing.. There's universal consensus these systems will save more lives. I'd argue it will save more lives than the users who abuse them. The famous case that comes to mind of the guy watching harry potter while the car (effectively slightly more advanced cruise control) drove into a semi.

 

[Donald Stanfield]

That system is only going to get better and if people assume responsibility and pay attention when using the system most of those accidents could be prevented. Tesla shifting the name from autopilot to "Full Self Driving" is the most damning indictment I think someone can make. Hence they started padding the (supervised).

In this very thread we have another guy who mentioned how he felt Rivian's system was dangerous. As I said previously, Rivian's Mobile-eye based solution (before their own rollout ) suddenly veered my R1T off an exit at highway speeds without any warning, but then I was still keeping my hands near the wheel and the brake press disabled the system in time.


The fact is these systems over time will save lives which people fail to mention:

National Safety Council:
"The study projects that the use of ADAS technologies could help avoid 249,400 fatalities and 14,138,000 nonfatal injuries from 2021 through 2050, in the most likely scenario. The most pessimistic scenario projects avoiding 152,100 fatalities and 8,673,000 nonfatal injuries. In comparison, the most optimistic scenario projects the avoidance of 298,300 fatalities and 16,814,000 nonfatal injuries. "

National Highway Traffic Safety Administration:
"NHTSA estimates that together these functions could potentially prevent up to nearly 600,000 crashes and about 1,300 fatalities annually when fully deployed"

This isn't even a Tesla apologist / vibes thing.. There's universal consensus these systems will save more lives. I'd argue it will save more lives than the users who abuse them. The famous case of the guy watching harry potter while their car's (effectively slightly more advanced cruise control) drives into a semi.

I agree that these systems are sound IF THE PEOPLE CONTINUE TO PAY ATTENTION. The Rivian Gen 2 system's eye tracking is pretty accurate. You must look at the road, and within a couple of seconds it will alert you and disengage. The problem I have with the Tesla system is for one, the name FSD is irresponsible.

Calling it full self drive, even if you have the little asterisk of supervised now, is a marketing decision that has gotten people killed to inflate a stock price and sell cars. Secondly, Tesla's system of checking if you're paying attention is defeatable, and was even more easily defeated in the past. This decision has also gotten people killed. By overstating FSD's capabilities and failing to put adequate safeguards in place, Tesla has acted irresponsibly. THAT is my issue with what Tesla is doing.

 

[sunydrm]

I agree that these systems are sound IF THE PEOPLE CONTINUE TO PAY ATTENTION. The Rivian Gen 2 system's eye tracking is pretty accurate. You must look at the road, and within a couple of seconds it will alert you and disengage.

Tesla's system is safer than that. It pulls over to the side of the road and shuts off if you are asleep or not paying attention. In the case of an accident it even pulls over and puts the hazards on.

In the case of the cameras being blinded it performs a minimum risk maneuver and will not shut off. FSD 14 can be instantly engaged from anywhere and it does not shut off anymore even in the most challenging conditions.

FSD 14 is also used for the new auto emergency braking which is a big leap over the past auto emergency braking which was already the best there is.

Tesla's system of checking if you're paying attention is defeatable, and was even more easily defeated in the past.

Rivian's system is defeatable in exactly the same way tesla's system is. You play a video of yourself driving and put it on loop in front of the driver camera. The difference is that tesla's system can actually drive for 50K miles with no safety critical intervention. You could actually defeat the system and not die. Rivian's system would shut off too often for this to have any purpose.

In the past before FSD had driver monitoring the system was harder to defeat. It involved buying a chip that you put into your steering wheel that would move the scroll wheel. Such chips were $150 and almost nobody had them

By comparison, a capacitive steering wheel like rivian has is much easier to defeat.

By overstating FSD's capabilities and failing to put adequate safeguards in place, Tesla has acted irresponsibly. THAT is my issue with what Tesla is doing.

Sounds like you're repeating the verdict from that $239 million dollar lawsuit or whatever it was. In that case it was autopilot, the driver was on the floor and he pressed the accelerator pedal where it disabled auto emergency braking. Tesla's logic is that human input while using ADAS will override auto emergency braking

 

[Donald Stanfield]

Tesla's system is safer than that. It pulls over to the side of the road and shuts off if you are asleep or not paying attention. In the case of an accident it even pulls over and puts the hazards on.

In the case of the cameras being blinded it performs a minimum risk maneuver and will not shut off. FSD 14 can be instantly engaged from anywhere and it does not shut off anymore even in the most challenging conditions.

FSD 14 is also used for the new auto emergency braking which is a big leap over the past auto emergency braking which was already the best there is.



Rivian's system is defeatable in exactly the same way tesla's system is. You play a video of yourself driving and put it on loop in front of the driver camera. The difference is that tesla's system can actually drive for 50K miles with no safety critical intervention. You could actually defeat the system and not die. Rivian's system would shut off too often for this to have any purpose.



Sounds like you're repeating the verdict from that $239 million dollar lawsuit or whatever it was. In that case it was autopilot, the driver was on the floor and he pressed the accelerator pedal where it disabled auto emergency braking. Tesla's logic is that human input while using ADAS will override auto emergency braking

I clearly said PREVIOUS systems. Tesla has only added the eye tracking after being sued for their irresponsibility. They had an easily defeated wheel feedback system before, which was way less realistic than videoing yourself driving, holding it up where your eyes are in order to defeat the system. If the trick is more work than paying attention, which is the case of the eye tracking workaround, people will pay attention. If you can tie anything weighted to the steering wheel people will do that.

Tesla should not be trusted as they have shown themselves to be irresponsible in this area.

 

[sunydrm]

I clearly said PREVIOUS systems. Tesla has only added the eye tracking after being sued for their irresponsibility.

Before tesla had eye tracking the system was harder to defeat. Not easier.

I think you're referring to using a weight on the steering wheel which does not work. Tesla detects constant torque and it gives you a strike and shuts off FSD. In fact Tesla is the only manufacturer with a strike system which causes you to lose access to a software you paid for.

By comparison if you do not have a torque based system (rivian capacitive steering wheel) it is very easy to defeat.

The reason tesla did not have driver monitoring is they did not have hands free. Tesla was pushing the limits of ADAS technology and they probably felt that hands free was both not safe and would make the car seem too autonomous (tesla was claiming their level 4 system is level 2). Now these systems are recognized as L2++ as more manufacturers have these

With FSD if you don't look at the road (or in the past tug the wheel) you get a strike and FSD shuts off. You used to lose access for 1 week after 5 strikes. Same with autopilot. I believe you would get 1 strike taken off for every week that you go without a strike, i.e. if you regain access with 4 strikes you lose it instantly after a 5th.

They had an easily defeated wheel feedback system before, which was way less realistic than videoing yourself driving, holding it up where your eyes are in order to defeat the system.

This is not how it works. You video yourself to replicate the look of the cabin camera and you mount the phone directly in front of the cabin camera. This would work with rivian too. This technique allows you to use FSD from the passenger's seat

Sponsored