Update: Cybertruck Full Self Driving (FSD) Option Now Costs $8K

[Newton]

So I am at work for 11 hours and my truck would just be sitting there. With FSD how will the cybertaxi work? Will I sign up for it and Tesla controls the hand off of my truck to someone and charges them and goes into an account for me? Want guarantees the truck is back when I need it to go home? What happens if I decide to leave early or have to run an errand I did not plan for.?

I see the income potential but confused on how the system will work? Any links to go read our YouTube video's to watch?

Dont expect your car to make u cash through the robotaxi service for a long time. At the very lowest estimate 4 years. unless your in a specific area of specific towns.
But that being said, if you plan to keep the vehicle for a while, over 7 years, it might be able to make u money on the side. If I had the cash to get the FSD option I probably would. but 7k now could go to much better use

Advertisement

 

[ajdelange]

Given that in the video Elon is "confident" that there will be a fleet of Tesla robotaxis on the road within the year or at most 13 months I am wondering when that video was shot. Were it yesterday I would say that his confidence is ill founded. FSD clearly has a long way to go before one would even think of turning it loose on the public highways for test let alone get it certified for transport of the public.

Based on a road trip yesterday I can't say that I think it has improved any in terms of what it can do in a year. But it can now show traffic lights which it couldn't a year ago. Smart Summon still does not work. It still slams on the brakes when it sees a structure over the road. It still tries to get off the freeway at the wrong ramp. It still makes my wife nervous as as the center on a gay football team.

But it still does relieve the driver of a lot of the work load on a long trip.

Now someone will come back and say "Wottsamotta U? It works fine for me!' We're talking about something to which the public would be exposed. It has to work for everyone 99.999% of the time. And I am not the only one who has been "unlucky".

If it did work it's easy to conjure up how it might work. Owners would have an app which they would use to communicate to a dispatch center the location, charge status and availabilities of their cars. People desiring a ride would have another app similar to the Lyft and Uber ones into which they would put information about whence they want to go and from where and when. Elon gave some hints as to how this might work in the video and indeed it may come to pass. But well past the date at which the first CT's hit the roadways.

 

[ldjessee]

The video was from last year's autonomy day, which was... April 2019, so his timeline would be by the end of 2020?

 

[Newton]

Now someone will come back and say "Wottsamotta U? It works fine for me!' We're talking about something to which the public would be exposed. It has to work for everyone 99.999% of the time. And I am not the only one who has been "unlucky".

Thats the KEY that people don't seem to realizeit need to work always because it dosnt have the human as a safety. the last 15% is MUCH much harder than the 1st 85%. and they aren't even at the 85%
Plus there will be unique cases where the car hasnt seen it and dosnt know what to do. like perhaps some1 carrying a mirror, or maybe a bunch of mylar baloons, or who knows, some1 walking with a long 2x4 it barely sees, etc.

 

[ajdelange]

The video was from last year's autonomy day, which was... April 2019, so his timeline would be by the end of 2020?

Thirteen months from then is right about now.

 

[ldjessee]

I do not think it has to be perfect. I do think it has to be significantly better than the average American driver before being approved to drive autonomously on the roads.

With the number of Tesla vehicles on the road increasing, the data they collect is increasing. This will help accelerate the training of the AI.

I have seen humans not handle things the AI handles. And that percentage will continue to grow.

I feel like it is unlikely for Tesla to start pressuring law makers by the end of this year to allow for autonomous driving in the US, but I think by the end of next year it will be very feasible.

 

[rr6013]

<Snip>

But it still does relieve the driver of a lot of the work load on a long trip.

Bingo! Ergo the core MPV...the rest is icing on the cake - someday

 

[rr6013]

I don’t get why so many complaints...it’s an option, just let it go, and be happy with the free driving assist.
I’ve driven both regular lane keep and speed control and there is no comparison to FSD.
My model S has FSD from last year ($6000) and the upgrades just keep coming on line. Really, what you are paying for at purchase is something you will receive in the future. I’m glad that I have FSD because my trade-in value for the CT will be that much more valuable.

Double check with Tesla before selling FSD with your used Tesla out into the open market. Tesla pulled back soft-ware options on a used Tesla sale. I don't think you " own" FSD to sell it. Just cross your T's and dot your I's before you add FSD into your resale price.

 

[rr6013]

Dont expect your car to make u cash through the robotaxi service for a long time. At the very lowest estimate 4 years. unless your in a specific area of specific towns.
But that being said, if you plan to keep the vehicle for a while, over 7 years, it might be able to make u money on the side. If I had the cash to get the FSD option I probably would. but 7k now could go to much better use

Second the motion " lower your expectations"!

Managed fleets in nationwide trucking and I have serious reservations " any" rent formula can "pay for itself" wrt covering Bank payment, maintenance and energy expenses in fleet usage... cashflow and depreciation curves will cross too early as competition drives pricing down.

I'm including in "competition" later Tesla models produced cheaper that flood into fleets as Uber history teaches.

 

[ajdelange]

I do not think it has to be perfect.

It doesn't. Governments seem to like 5 nines. Based on the number of errors that required intervention I experienced in driving with it for 300 mi on Saturday I would say they are currently at perhaps 2.

I do think it has to be significantly better than the average American driver before being approved to drive autonomously on the roads.

The government may approve it at the 5 nines level but there is more to it than that. Every third word in the owner's manual description for the system seems to be "beta" and the remaining are repeated caveats that the driver has responsibility for the vehicle at all times when autopilot is on. Yet a guy out in California is driven into a barrier by it and killed. Early witnesses at the scene find he was watching a movie on a DVD player at the time of the crash. But this does not deter the guy's family from suing Tesla. The point being that Tesla will want as many 9's as it takes to get the expected value of litigation related costs below the expected value of profits from sales of this tech. Raising the price of the package, as they have just done, is, of course one way but getting an extra 9 or 2 is another.

With the number of Tesla vehicles on the road increasing, the data they collect is increasing. This will help accelerate the training of the AI.

As the rate at which Tesla is putting vehicles on the road is increasing exponentially using that as a criterion today's autopilot should be at least noticeably better than last year's. It isn't. If anything it seems worse (but I am a sample of 1).

I feel like it is unlikely for Tesla to start pressuring law makers by the end of this year to allow for autonomous driving in the US, but I think by the end of next year it will be very feasible.

I have found over the years that there are two groups of people who have unrealistic expectations regarding AI. It is not surprising that those who know nothing about it at all are found in one of these. But the other is the ones who know the most about it, that is, the experts. I have been listening to them for years tell me that it's just a year or two away. Plus ça change, I suppose.

Have you actually ridden in a Tesla vehicle when autopilot was in use? If not try to do that. I think you will be brought to a more reasonable perspective. While I think everyone should form his own opinion there is a poll somewhere on here in which respondents listed their expectations. You might find that interesting.

 

[ajdelange]

Bingo! Ergo the core MPV...the rest is icing on the cake - someday

Not for someone who is hoping to use it as a robotaxi.

PS; What does MPV stand for?

 

[rr6013]

Not for someone who is hoping to use it as a robotaxi.

PS; What does MPV stand for?

Typo - MVP
Minimum Viable Product

 

[ldjessee]

Have you actually ridden in a Tesla vehicle when autopilot was in use? If not try to do that. I think you will be brought to a more reasonable perspective. While I think everyone should form his own opinion there is a poll somewhere on here in which respondents listed their expectations. You might find that interesting.

No, not yet ridden in a Tesla. But I have ridden in several helicopters and seen what is possible with automated piloting hardware/software 20 years ago. Sure things do not advance as fast as we would like or sometimes expect, but I think automated driving is not as hard as people think it is, just requires time and effort.

This interview is interesting, though it is someone who has a vested interest in making automation look possible.

 

[ajdelange]

No, not yet ridden in a Tesla. But I have ridden in several helicopters and seen what is possible with automated piloting hardware/software 20 years ago. Sure things do not advance as fast as we would like or sometimes expect, but I think automated driving is not as hard as people think it is, just requires time and effort.

Well that's where you diverge from the way most people see it. I haven't concerned myself with aviation auropilots since I had a servo transistor go TU on me during an approach to San Jose (I think it was) so I went and did a quick search on autoland. The one article I read described a situation quite parallel with what you will find in cars i.e. the pilots are required to be hypervigilant to the point that many consider the workload greater than that of doing the landing themselves. And the job is much easier with an aircraft in that you have cm accuracy position information from a pseudolite at the airport (I assume they don't allow automatic Cat III landings just anywhere). And you don't have mothers pushing a pram out from between parked cars on a runway. Runways have standard markings anywhere in the world and the airspace uses the same navaids. Before an approach is authorized it is certified. Do look at aviation but look at it from the point of view of how the US highway system would have to be modified to make it as secure as the airspace system. On Saturday I was driving in DC with the sun right off the nose at an oblique angle. I couldn't see the roadmarkings and neither could the cameras. This is a problem that has to be solved, among many others, before robo taxiing will be possible.

etc.

 

[Newton]

I was listening to something, I think "Our Ludicrous Future" podcast and the everyday astronaut guy was saying something like... We have yet to even automate rail, we still have a conductor and this is only 1 degree of freedom. Seems similar to planes, we still need a pilot waitching over it and these also have pre-routed paths (correct me if im wrong).