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!

 

[aataskin]

May even be 2 prototypes (2 seater and a van/5+ type) since they are relatively easy to build.

Could be. I just don't understand the excitement over how a robotaxi may look like without the company demonstrating first that it has the required technology for a robotaxi. L2 is not good enough, and a prototype with L2, no matter how shocking it may look like on the outside, is in reality pretty pointless.
Unlike the CT, this presentation is not meant to be about a new form factor, but rather the technology that would make the Tesla robotaxi something more than a glorified car rental service.

 

[stopcrazypp]

Could be. I just don't understand the excitement over how a robotaxi may look like without the company demonstrating first that it has the required technology for a robotaxi. L2 is not good enough, and a prototype with L2, no matter how shocking it may look like on the outside, is in reality pretty pointless.
Unlike the CT, this presentation is not meant to be about a new form factor, but rather the technology that would make the Tesla robotaxi something more than a glorified car rental service.

For Waymo, having the Firefly (the small two seater similar to the concept here) was what drove excitement. They demonstrated the car with a short drive in a residential neighborhood with a blind man in it. At the time they had Prius prototypes too, but no one would be excited about those. Tesla can totally do a similar demo drive in the prototype also.

As for expecting demonstrating "required technology," there is no universal accepted minimal standard for a "robotaxi" so that is pretty hard to define. Door-to-door L2 and L4 under test is sufficiently similar, that most people can't tell the difference. They can simply remove the steering wheel and pedals, claim they are operating in L4 mode (if unveiled in California, submit a disengagement report), pick up a passenger, and it would be technically true.

If you are talking about slapping on some lidar sensors, putting aside it is not proven lidar is "required" for L4, that will do nothing to demonstrate more L4 readiness, as the software is what matters. There are many cars out there loaded to the gills with sensors (including lidar), but far less capable than a Tesla with FSDS.

 

[enemji]

Could be. I just don't understand the excitement over how a robotaxi may look like without the company demonstrating first that it has the required technology for a robotaxi. L2 is not good enough, and a prototype with L2, no matter how shocking it may look like on the outside, is in reality pretty pointless.
Unlike the CT, this presentation is not meant to be about a new form factor, but rather the technology that would make the Tesla robotaxi something more than a glorified car rental service.

ODD applies.

I am pretty sure this RT will be available only on certain paths only. Ones where there is revenue such as RT to the Airport.
Delivering that will be easy as NN can be trained for those specific routes very easily.

 

[aataskin]

As for expecting demonstrating "required technology," there is no universal accepted minimal standard for a "robotaxi" so that is pretty hard to define.

I think it's well understood that a robotaxi, just like a regular taxi, is something that takes you from point A to point B without you paying attention to the road, and is liable if it gets into an accident. Unlike FSD which is a unique name so Tesla can claim that people misinterpreted what it means, the word "taxi" is very well defined.

Door-to-door L2 and L4 under test is sufficiently similar, that most people can't tell the difference.

Not sure what you mean. L4 is something where I don't need to monitor the road. L2 is where I am the one responsible if the car gets into an accident. The difference is pretty obvious.

If you are talking about slapping on some lidar sensors, putting aside it is not proven lidar is "required" for L4, that will do nothing to demonstrate more L4 readiness, as the software is what matters. There are many cars out there loaded to the gills with sensors (including lidar), but far less capable than a Tesla with FSDS.

But none of those cars claim to be robotaxis.

 

[stopcrazypp]

I think it's well understood that a robotaxi, just like a regular taxi, is something that takes you from point A to point B without you paying attention to the road, and is liable if it gets into an accident. Unlike FSD which is a unique name so Tesla can claim that people misinterpreted what it means, the word "taxi" is very well defined.

But what is the minimal standard for one? For example what if it only does a trip in a small area (and can even be a closed or low traffic area) on predefined routes? FSDS can totally be made to do that today.

Not sure what you mean. L4 is something where I don't need to monitor the road. L2 is where I am the one responsible if the car gets into an accident. The difference is pretty obvious.

I'm talking about L4 UNDER TEST (meaning with safety drivers to monitor the car as required in California in early stages). There is a very thin line between those two.

But none of those cars claim to be robotaxis.

Fair enough, but doesn't change the point is that it is not proven that lidar is "required" nor does lidar automatically mean suddenly the car is "more ready".

 

[DrChaos]

Almost positive there will be a prototype shown on 8/8. CT, M3, etc. have all had a car hand built to show upon unveiling.

And other than the M3 (which is a long time ago with less Elon ego) none of those meant that production engineering was anywhere near complete or production installation had even started.

 

[Daniel in SD]

https://twitter.com/x/status/1801980372823187707

Plot twist: The robotaxi will be based on roadster platform. Flying is infinitely easier to automate than driving.

 

[aataskin]

But what is the minimal standard for one? For example what if it only does a trip in a small area (and can even be a closed or low traffic area) on predefined routes? FSDS can totally be made to do that today.

Robotaxi Limited, or FSD Predefined, something like that. I hope you are right and this is what they go for because then we will all be able to clearly see what limits Tesla themselves place on the usefulness of their FSD (since they will be the ones liable in case of an accident).

 

[aataskin]

https://twitter.com/x/status/1801980372823187707

Plot twist: The robotaxi will be based on roadster platform. Flying is infinitely easier to automate than driving.

Not if you stick to Vision only. But if you allow GPS, gyroscope and equipment for ILS then yes, with all this additional equipment then yes, it should be infinitely easier to achieve than driving. Infinitely being the key word.

 

[Daniel in SD]

Something like this. A flying robotaxi would also explain the two seat rumors which otherwise make zero sense.

 

[EvilNuff]

Could be. I just don't understand the excitement over how a robotaxi may look like without the company demonstrating first that it has the required technology for a robotaxi. L2 is not good enough, and a prototype with L2, no matter how shocking it may look like on the outside, is in reality pretty pointless.
Unlike the CT, this presentation is not meant to be about a new form factor, but rather the technology that would make the Tesla robotaxi something more than a glorified car rental service.

Wow, just when I thought sanity had fully departed this forum you posted this and restored my faith in humanity. Absolutely 100% dead on completely accurate. I will go one step farther and point out (again) that full robotaxi (which is level 5 not 2, not 3, not 4, 5 autonomy) as promised is simply not possible (ever) with current tesla hardware.

 

[DrChaos]

Wow, just when I thought sanity had fully departed this forum you posted this and restored my faith in humanity. Absolutely 100% dead on completely accurate. I will go one step farther and point out (again) that full robotaxi (which is level 5 not 2, not 3, not 4, 5 autonomy) as promised is simply not possible (ever) with current tesla hardware.

let's remember the promises on the Battery Day, and DOJO, solar roof, and all that and then the reality of what came afterwards

In a nutshell the in house stuff is mostly a bust and they're still buying batteries 95% from other suppliers, and they're buying ML systems 95% from other suppliers. Energy storage is decent, Semi sales are minuscule, and solar a failure.

they can make a presentation which will fool the naive, but the experts have seen it all before. The one advantage is if unsupervised training of humans "wild in the field" can make for a useful system. But it may be like the LLMs----training on all unsupervised text in the wild makes for systems which have numerous glitches though flashes of brilliance. The feedback and tuning and 'alignment' to make them safe also gives them a lobotomy.

Will driving be the same? And even with a tech demo there is a tremendous amount of gnarly labor needed to make for revenue taxis services. Waymo has tech Tesla will get to only in many years and yet they can't scale and expand profitably quickly yet.

 

[aronth5]

Could be. I just don't understand the excitement over how a robotaxi may look like without the company demonstrating first that it has the required technology for a robotaxi. L2 is not good enough, and a prototype with L2, no matter how shocking it may look like on the outside, is in reality pretty pointless.
Unlike the CT, this presentation is not meant to be about a new form factor, but rather the technology that would make the Tesla robotaxi something more than a glorified car rental service.

First, current hardware will support L3 if Tesla decides to enable that as part of their overall FSD solution. Will they? Probably not until another car maker releases a good L3 solution that will actually sell not the silly Mercedes system. Easily well over half of my drives now would support a Robotaxi level solution.

Second, I expect 8/8 to primarily focus on the unveiling of the new Robotaxi car without pedals and steering wheel. Tesla won't answer many of the questions about the car nor detailed information about the actual Robotaxi service but that is typical of all their vehicle unveilings. Tesla won't get into specific dates or information on the initial pilot program. So the car will be exciting since it will be different than any vehicle on the road today.

 

[aataskin]

First, current hardware will support L3 if Tesla decides to enable that as part of their overall FSD solution. Will they? Probably not until another car maker releases a good L3 solution that will actually sell not the silly Mercedes system. Easily well over half of my drives now would support a Robotaxi level solution.

Well over half is not good enough. Even 99% is not good enough. And, while we are at it, L3 is not good enough.

Second, I expect 8/8 to primarily focus on the unveiling of the new Robotaxi car without pedals and steering wheel. Tesla won't answer many of the questions about the car nor detailed information about the actual Robotaxi service but that is typical of all their vehicle unveilings. Tesla won't get into specific dates or information on the initial pilot program. So the car will be exciting since it will be different than any vehicle on the road today.

For little kids, yes, it will be.

 

[uscbucsfan]

Well over half is not good enough. Even 99% is not good enough. And, while we are at it, L3 is not good enough.

No solution will ever be 100%, Waymo, Cruise, Zoox, Tesla, etc.

That's just science, no driver or vehicle is 100% reliable, ever.

99% is perfectly sufficient with the ability to remote operate as we see with Waymo (which I doubt is even 99%).

 

[diplomat33]

99% is perfectly sufficient with the ability to remote operate as we see with Waymo (which I doubt is even 99%).

Waymo is 100x better than 99%. Based on their collision rate, Waymo is at 99.997%.

I don't think 99% is good enough. That would mean 1 mile out 100 has an intervention. And even if you are saying that intervention is not safety critical and can be handled by a remote operator, that would be a lot of remote interventions for any fleet of any decent size. For example, it would be 10,000 remote interventions every 1M miles. If your robotaxis were doing 1M miles every month, that would mean 10,000 remote interventions a month. That would not be scalable. I seriously doubt that Waymo needs 1000s of remote interventions per month. You need more "9s". Waymo is at 99.997%. So I would say you probably need at least 99.99% to start driverless.

 

[kabin]

Hopefully the 9's are well defined otherwise Tesla will game it. There are a few ways to interpret it. One mile out of 1000 miles could be a slew of interventions each occurring in short distances that add up to 1 mile. Or 1 intervention out of 1000 FSD decisions.

 

[uscbucsfan]

Waymo is 100x better than 99%. Based on their collision rate, Waymo is at 99.997%.

I don't think 99% is good enough. That would mean 1 mile out 100 has an intervention. And even if you are saying that intervention is not safety critical and can be handled by a remote operator, that would be a lot of remote interventions for any fleet of any decent size. For example, it would be 10,000 remote interventions every 1M miles. If your robotaxis were doing 1M miles every month, that would mean 10,000 remote interventions a month. That would not be scalable. I seriously doubt that Waymo needs 1000s of remote interventions per month. You need more "9s". Waymo is at 99.997%. So I would say you probably need at least 99.99% to start driverless.

I think Waymo intervenes far more than you think.

I can't find any published data, but Cruise was intervening every 5 miles on average.

 

[aataskin]

I think Waymo intervenes far more than you think.

I can't find any published data, but Cruise was intervening every 5 miles on average.

And where's Cruise?

 

[uscbucsfan]

And where's Cruise?

Only in Arizona now.