RJ: Rivian ADAS will expand to Tesla style FSD - full self driving on highway and roads (from driveway to destination)

[ndmiller]

Tesla FSD will never be 100% w vision only, you gotta have the lidar as well - I argue my friend who's a Tesla fangirl all the time about this.

100% pure speculation with no backing but your opinion. I drive myself around all the time with only vision and no lidar and so does everyone in my family.

FWIW I love these nonsense threads where everyone pretend sthey actually know anything about the topic. Meanwhile, I get to ignore another 3-4 people making the forum a better experience.

Sponsored

 

[captainjp]

100% pure speculation with no backing but your opinion. I drive myself around all the time with only vision and no lidar and so does everyone in my family.

FWIW I love these nonsense threads where everyone pretend sthey actually know anything about the topic. Meanwhile, I get to ignore another 3-4 people making the forum a better experience.

Much of the same can be said about the counter argument

 

[EVtowing]

I drive myself around all the time with only vision and no lidar and so does everyone in my family.

Elon has said multiple times something to the effect of “Since humans navigate the world with eyes and a brain cars should be able to do the same with cameras and a computer” as his reason for not using lidar, radar, and ultrasonics and for promising, every year for 6+ years, that Tesla FSD will be capable of Level 4/5 operation “this year”. He’s been wrong every time. See https://electrek.co/2025/03/23/tesla-full-self-driving-stagnating-after-elon-exponential/

Elon’s argument is flawed. He thinks that current technology is equivalent to human biology. In some ways the tech is better, and in other ways it is worse. In addition, our roads and signage are designed for humans, not for cameras connected to a computer.

So yes, we humans drive our cars with just vision and visual processing in our brain (but we also use our hearing) that then in turn sends electrical impulses to our arms and legs to control our cars. No lidar required.

 

[Rivianero]

While I appreciate basic ADAS features like dynamics cruise control and lane keeping, I recognize that the development investment required for more advanced automation significantly impacts Rivian's profitability, just as the sensors and compute required adds to overall price.

Cool stuff, but I would rather have a solvent company, lower vehicle costs, and (frankly) other less costly standard features like reliable 12v batteries, key fobs, spare tires, text integration and such.

I'm not presently in the market to buy another vehicle, but Rivian should reconsider what useful things they are omitting from the vehicles in order to chase after whiz-bang features that never quite work as promised.

I'm sure that full self-driving is coming... some decade... really soon now. In the meantime, all manufacturers seem to be chasing after it as a must-have gimmick to the exclusion of other, actually attainable and valuable features. It seems delusional that any buyer would believe promises of true self-driving since no car builder has ever truly delivered on it. We can talk again once that is accomplished, but right now you're just buying a whimsical promise.

 

[Donald Stanfield]

Elon has said multiple times something to the effect of “Since humans navigate the world with eyes and a brain cars should be able to do the same with cameras and a computer” as his reason for not using lidar, radar, and ultrasonics and for promising, every year for 6+ years, that Tesla FSD will be capable of Level 4/5 operation “this year”. He’s been wrong every time. See https://electrek.co/2025/03/23/tesla-full-self-driving-stagnating-after-elon-exponential/

Elon’s argument is flawed. He thinks that current technology is equivalent to human biology. In some ways the tech is better, and in other ways it is worse. In addition, our roads and signage are designed for humans, not for cameras connected to a computer.

So yes, we humans drive our cars with just vision and visual processing in our brain (but we also use our hearing) that then in turn sends electrical impulses to our arms and legs to control our cars. No lidar required.

The biggest thing I don't get with this line of thinking is that humans aren't perfect at driving, so why would you want to give the vehicle all the limitations of human vision? The better solution is to do human+ vision. More views, radar, and lidar to see things that humans cannot. See through the weather, glare, and let the system catch objects that are difficult for a human to see in adverse conditions like the dark or during a rain/snow storm.

We know those are areas that human vision struggles. Anyone who has driven in a whiteout or experienced how dark it is on a back country road at night can tell you that human vision has obvious limits. We can eliminate those limits, so if the same AI he uses with human vision also incorporates extra sensors, we will achieve software that drives better than a human ever could. A KITT-type car would be a reality minus the jumping, rockets, bullets, and stuff.

 

[pamalabama]

The biggest thing I don't get with this line of thinking is that humans aren't perfect at driving, so why would you want to give the vehicle all the limitations of human vision? The better solution is to do human+ vision. More views, radar, and lidar to see things that humans cannot. See through the weather, glare, and let the system catch objects that are difficult for a human to see in adverse conditions like the dark or during a rain/snow storm.

We know those are areas that human vision struggles. Anyone who has driven in a whiteout or experienced how dark it is on a back country road at night can tell you that human vision has obvious limits. We can eliminate those limits, so if the same AI he uses with human vision also incorporates extra sensors, we will achieve software that drives better than a human ever could. A KITT-type car would be a reality minus the jumping, rockets, bullets, and stuff.

Your logic is that cameras have the same limitations as human vision. They don't.

Cameras see better in the dark then our eyes do and they have higher dynamic range.

Cameras can see into the sun without having glare, ghosting or afterimages

 

[pamalabama]

Elon has said multiple times something to the effect of “Since humans navigate the world with eyes and a brain cars should be able to do the same with cameras and a computer” as his reason for not using lidar, radar, and ultrasonics and for promising, every year for 6+ years, that Tesla FSD will be capable of Level 4/5 operation “this year”. He’s been wrong every time. See https://electrek.co/2025/03/23/tesla-full-self-driving-stagnating-after-elon-exponential/

Elon’s argument is flawed. He thinks that current technology is equivalent to human biology. In some ways the tech is better, and in other ways it is worse. In addition, our roads and signage are designed for humans, not for cameras connected to a computer.

So yes, we humans drive our cars with just vision and visual processing in our brain (but we also use our hearing) that then in turn sends electrical impulses to our arms and legs to control our cars. No lidar required.

If you were not ignorant about FSD you would know that FSD has exponentially improved with each version. The reason that the miles per disengagement are the same is because tesla does not fix easy issues. Tesla does not read school zone signs, so you can go through a school zone every single day and have a disengagment.

you can also NOT go through a school zone and go more than one month without ever touching the steering wheel or pedals

Tesla hasn't had a significant issue in years that was caused by using cameras only

80% of tesla's issues can be solved without a single change to FSD just by using better maps

Robotaxi launch is in June. Assuming it is summer and there is no school zone or school bus in sight and FSD has "perfect" map data, I could see how elon can say that interventions are exceedingly rare

 

[pamalabama]

It would be interesting if Rivian can even do much in the way of autonomy.

Most cars do not have a steering column control module that is capable of autonomy because it is a safety risk. Having a car perform a signicant and fast steering wheel move at high speed can cause a car to flip over. With Gen 1, Rivian was using a Hyundai part which seems to be almost incapable of even turning the steering wheel more than 90 degrees, not allowing you to take even moderate curves.

It takes a lot of steering wheel turning to make a 90 degree turn, make a uturn, or drive a switchback.

 

[KootenayEV]

Your logic is that cameras have the same limitations as human vision. They don't.

Cameras see better in the dark then our eyes do and they have higher dynamic range.

Cameras can see into the sun without having glare, ghosting or afterimages

Uh, no they don't! (Source: hobby photographer for 20 years)

 

[Ralph]

If you were not ignorant about FSD you would know that FSD has exponentially improved with each version. The reason that the miles per disengagement are the same is because tesla does not fix easy issues. Tesla does not read school zone signs, so you can go through a school zone every single day and have a disengagment.

you can also NOT go through a school zone and go more than one month without ever touching the steering wheel or pedals

Tesla hasn't had a significant issue in years that was caused by using cameras only

80% of tesla's issues can be solved without a single change to FSD just by using better maps

Robotaxi launch is in June. Assuming it is summer and there is no school zone or school bus in sight and FSD has "perfect" map data, I could see how elon can say that interventions are exceedingly rare

I don't know anything about FSD's ability, but I'm very skeptical about Rivian's announcement unless they are perhaps going to integrate another vendor into their hardware and software stack.

But why wouldn't Tesla address the "easy issue" of school zones, etc.? Not addressing something that is "easy" doesn't make any sense to me.

Perhaps I am not understanding?

 

[pamalabama]

I don't know anything about FSD's ability, but I'm very skeptical about Rivian's announcement unless they are perhaps going to integrate another vendor into their hardware and software stack.

But why wouldn't Tesla address the "easy issue" of school zones, etc.? Not addressing something that is "easy" doesn't make any sense to me.

Perhaps I am not understanding?

There is a patent owned by mobileye on reading road signs. So tesla just drove without reading most road signs

Tesla's goal is to have the car drive in every scenario possible. The car gets better without the easy issues being fixed

 

[pamalabama]

Uh, no they don't! (Source: hobby photographer for 20 years)

Hobby photographer is not the same thing as self driving.

The cameras used for automotive are not built to produce beautiful images. They are built to capture the most dynamic range. The automotive sensors are HDR with 2 sizes of photosites, backside illumination AND dual gain. The 3 most common methods to increase dynamic range all in one sensor.

 

[Ralph]

There is a patent owned by mobileye on reading road signs. So tesla just drove without reading most road signs

If true, seems kind of silly for Tesla not to pay up if they are going to compete in their chosen market.

 

[pamalabama]

If true, seems kind of silly for Tesla not to pay up if they are going to compete in their chosen market.

Tesla open sourced their patents for a reason. So they can freely use other patents they probably need. It just so happens that mobileye does not make an electric vehicle.

if I'm not mistaken 360 view is owned by nissan? And the leaf is such a poor electric vehicle it may not be using any tesla patents like everyone else does

plus tesla has no incentive to read road signs until they need to launch robotaxi. The car drives 99% fine without reading them now

 

[pamalabama]

FSD does the "hard stuff" very well. Driving in heavy rain, snow covered roads where you can't see any markings, fog, narrow mountain roads with just dirt, trails, etc.

In those limited scenarios it seems to see better than I can.

partially because FSD uses satellite images to assist in driving so it knows what's ahead during blind curves

FSD also sees solutions to routes you would never figure out on your own



Look at this. Insane

Sponsored