Make your robotaxi predictions for the 8/8 reveal

So Elon says that Tesla will reveal a dedicated robotaxi vehicle on 8/8. What do you think we will see? Will it look like this concept art or something else?

I will say that while this concept drawing looks super cool, I am a bit skeptical if it is practical as a robotaxi. It looks to only have 2 seats which would be fine for 1-2 people who need a ride but would not work for more than 2 people. I feel like that would limit the robotaxis value for a lot of people. Also, it would likely need a steering wheel and pedals for regulatory reasons even if Tesla did achieve eyes-off capability.

So I think this is concept art for a hypothetical 2 seater, cheap Tesla, not a robotaxi.

Could the robotaxi look more like this concept art but smaller? It could look a bit more like say the Zoox vehicle or the Cruise Origin, more futuristic box like shape IMO and seat 5-6 people.

Or maybe the robotaxi will look more like the "model 2" concept:

Other questions:
- Will the robotaxis be available to own by individuals as a personal car or will it strictly be owned by Tesla and only used in a ride-hailing network?
- What will cost be?
- Will it have upgraded hardware? Radar? Lidar? additional compute?
- Will Elon reveal any details on how the ride-hailing network will work?

Thoughts? Let the fun speculation begin!

 

[uscbucsfan]

That is what is happening right now

When?

Seriously, though, the RT will not be based on the 3/Y unless Tesla changes direction, again. It's a completely new platform.

 

[enemji]

When?

Seriously, though, the RT will not be based on the 3/Y unless Tesla changes direction, again. It's a completely new platform.

It will be based on 3/Y. we can wait for 8/8 and see who is right.

Or better yet this is what I believe would really work

 

[uscbucsfan]

It will be based on 3/Y. we can wait for 8/8 and see who is right.

View attachment 1052064

View attachment 1052068

Or better yet this is what I believe would really work

View attachment 1052066

Why do you think that? You think Elon, Franz, and Tesla are all just misleading for the surprise on 8/8? They've said it was a new design and the first unboxed car.

Almost 0 chance.

 

[enemji]

Why do you think that? You think Elon, Franz, and Tesla are all just misleading for the surprise on 8/8? They've said it was a new design and the first unboxed car.

Almost 0 chance.

Do they have the capacity to cast full bodies they are promising? I don’t think so. At least not for another 24 months

 

[uscbucsfan]

Do they have the capacity to cast full bodies they are promising? I don’t think so. At least not for another 24 months

A new gigapress was recently delivered to Austin. 8/8 is unveiling. Who knows when the launch is.

It's almost a guarantee they aren't basing it off the MY. Most of the plans from Franz, the Elon Biography, photos from the Tesla video indicate it's a 2-door, 2 seater car, but regardless, Elon/Tesla stated there are new cars being released on the 3/Y platform, using their lines and the RT, which is unboxed and a new design.

 

[JulienW]

A new gigapress was recently delivered to Austin. 8/8 is unveiling. Who knows when the launch is....

Yeah whatever is shown will almost certainly be a hand built prototype and NOT close to a production or even a preproduction car. Production will likely start "in about 6 months" Elon time after.

 

[uscbucsfan]

Yeah whatever is shown will almost certainly be a hand built prototype and NOT close to a production or even a preproduction car. Production will likely start "in about 6 months" Elon time after.

Right.

 

[JulienW]

Also I think the 2 seater will be the most efficient for a RT since almost all taxi rides are 2 or less. Could be the platform we be modular enough to allow a van type 6 seater with larger battery to be built and added to the RT fleet at a later date for groups or ADA access.

 

[enemji]

A new gigapress was recently delivered to Austin. 8/8 is unveiling. Who knows when the launch is.

That is good to know. If that is the case, then it is possible. However I would still say the overall dimensions and camera placement locations would be the same as the Model3/Y in order to leverage the existing NN fine tuned for FSDs

 

[uscbucsfan]

That is good to know. If that is the case, then it is possible. However I would still say the overall dimensions and camera placement locations would be the same as the Model3/Y in order to leverage the existing NN fine tuned for FSDs

They are already testing new camera locations. There's an image on this site somewhere.

 

[JulienW]

They are already testing new camera locations. There's an image on this site somewhere.

I'd bet HW5 too.

 

[enemji]

They are already testing new camera locations. There's an image on this site somewhere.

Testing is fine, but having the same perspective on which the current NN is trained is important. One cannot just put cameras anywhere they like., unless they want to retrain it all over again for every little scenario

 

[uscbucsfan]

Testing is fine, but having the same perspective on which the current NN is trained is important. One cannot just put cameras anywhere they like., unless they want to retrain it all over again for every little scenario

I'm sure Telsa will heed your advice.

 

[enemji]

I'm sure Telsa will heed your advice.

They better if they want to enter the market quickly

 

[DrChaos]

Yeah whatever is shown will almost certainly be a hand built prototype and NOT close to a production or even a preproduction car. Production will likely start "in about 6 months" Elon time after.

The CT production turned out to be very different underneath than the CT demo, because the demo couldn't be manufactured or pass crash.

 

[DrChaos]

Testing is fine, but having the same perspective on which the current NN is trained is important. One cannot just put cameras anywhere they like., unless they want to retrain it all over again for every little scenario

They have some 3d normalization now into a shared space so they aren't all independent views, so its not as bad as you think, but new cameras will still require tuning.

 

[JulienW]

They have some 3d normalization now into a shared space so they aren't all independent views, so its not as bad as you think, but new cameras will still require tuning.

As I said earlier I bet it is HW5 and the cameras are probably higher resolution and maybe even a higher bit depth for a wider color space. Also I bet they just take v12 as a base and then a custom version for the RT will be trained on the HW5 suite and NOT the HDW3 suite like V12 is.

 

[DrChaos]

As I said earlier I bet it is HW5 and the cameras are probably higher resolution and maybe even a higher bit depth for a wider color space. Also I bet they just take v12 as a base and then a custom version for the RT will be trained on the HW5 suite and NOT the HDW3 suite like V12 is.

The problem with the RT self-driving is that there won't be extensive human driving with the same hardware to train it.

The original goal, before Musk altered the deal, of making a new low cost platform for both human and robotic driving was the right one. Get manufacturing cheap, sell tons of human driven cars, which will train the ML models so that you can ease into a RT on a similar platform, one with the same sensors and dynamics so that there is minimal gap in the models when adapting to RT. RT autonomous driving is so risky and will be scrutinized so much that they need to minimize all risks. Also, for RTs to be profitable the actual hardware needs to be cheap (Waymo problem) and that happens on the very highest volume production line, the one pumping out China-competitive cars for the masses. Human driven cars, the ones that you can sell for money today.

Well, Elon just blew that strategy up because of short term profitability worries or adderall.

Now they're going to have the Waymo problem of needing to invest in lots of employees driving non-revenue producing RTs to gather training data manufactured on a bespoke line that will have no revenue incoming at all until RTs are perfected which is so many years away. Theyre going to have trouble scaling up manufacturing simultaneously to having to scale up RT revenue, and both will be rocky.

My prediction unfortunately: there will never be a significant RT revenue from Tesla for 10 years

 

[enemji]

The problem with the RT self-driving is that there won't be extensive human driving with the same hardware to train it.

The original goal, before Musk altered the deal, of making a new low cost platform for both human and robotic driving was the right one. Get manufacturing cheap, sell tons of human driven cars, which will train the ML models so that you can ease into a RT on a similar platform, one with the same sensors and dynamics so that there is minimal gap in the models when adapting to RT.

100% on point. I am glad you took the time to provide that perspective of being fast & cheap, which is why I pushed for RT based on the Model3/Y platform or at least the same camera locations at the minimum with close enough body size. You go about retraining with new set of HW with cameras and other sensors(?), and it will neither be fast nor cheap.

 

[DrChaos]

100% on point. I am glad you took the time to provide that perspective of being fast & cheap, which is why I pushed for RT based on the Model3/Y platform or at least the same camera locations at the minimum with close enough body size. You go about retraining with new set of HW with cameras and other sensors(?), and it will neither be fast nor cheap.

I really do wonder if the RT hardware platform will ever really take off, or that will be abandoned a few years in. I agree that it seems more feasible that once they have a mass production lower-cost car ready they'll try to make robotaxis off of it as it's easier to guarantee the machine learning model will work well as there are millions of FSD customers to train and test it.

The thing though is that a RT needs significant interior and door reconfiguration (need fully automatic doors) which might not be really feasible being based on a Model 3 Minus. Like removing the b-pillars entirely changes the structure and engineering design & crash issues.

The fully new design, which was supposed to be for human cars as well until Musk just killed it, was intended to have very low long term manufacturing costs which was why they were doing it in the first place. TBH it seems like Tesla was being run well until Elon came back from his Twitter phase and started swinging his dick around.