Tesla Should Have Early Access Release of Feature Complete Full Self-Driving by the End of 2019

At about 1 hour and 6 minutes (about 5 minutes into the Webcast) of this video of the Tesla Q3 Earnings call, Elon Musk said that Tesla should have early access for a feature-complete full self-driving system by the end of the year. This will be a tight delivery of a feature-complete full self-driving system by the end of the year.

Elon’s Self Driving Major Levels
Self Driving Level 1. Feature complete means supervised driving at all speeds but not necessarily all corner cases. Human intervention would not be needed in most cases. It will need supervision and some intervention in some cases.
Self Driving Level 2. Tesla believes the cars will be safe enough without supervision
Self Driving Level 3. Regulators believes it will be safe to drive without supervision

Tesla will have version 3 of the solar power roof. This will be officially launched tomorrow. Elon Musk this roof will be ready for the big time.

At about 1:20 (19 minutes into the webcast), the video has questions and answers.

Elon believes Tesla’s energy (solar plus battery) business will be as big or bigger than Tesla’s automotive business.

Tesla Energy was stripped of technical resources due to the problems with the Tesla Model 3 ramp. These issues are overcome and resources are going to Tesla Solar and Storage. This will have “really crazy growth” for a long time.

Elon drove the Model Y release candidate. In Elon’s opinion the Model Y will outsell Model 3, Model X and Model S combined. Launching Model Y will not impact the production of Model 3. Model Y will have higher margin than Model 3.

Solar helps increase the value of a home by about 4% according to Zillow.
Solar and Powerwall will be especially valuable in California where there will be long term power problems.

Tesla has a generic template for getting solar roofs approved. This is an innovation in the bureaucracy space. Tesla’s cost of acquisition are only 25% of other solar companies because they only have online ordering and no advertising.

Tesla will continue to sell self-driving as a package with their cars. The current focus for their self-driving work is on the handling of traffic lights and small windy roads.

The Robotaxi capability should be very reliable by the end of next year. It will be feature complete this year.
Changing from non-robotaxi to robotaxi will be the biggest step-change increase in asset value in history for a product.

Full-self driving (FSD)price should increase slowly over time as its capabilities improve.

Tesla gross margin is higher because there is about $500 million of unrecognized FSD revenue.

Deepscale was a tiny AI company with 12 people. The goal is to slightly increase the speed of FSD development.

Tesla should hit volume production in the China factory in a few months.
The second building at the Shanghai factory is for battery and module production.
The China factory will be expanded to handle Model Y production.

A Feature complete self-driving Tesla car will be able to drive from your home to work mostly likely without intervention. It will still be supervised.

There will be ramping inefficiencies during Q4 for the China factory production ramp.

Elon believes the Tesla Pickup truck will be their best product ever. Details of the Tesla pickup truck will be available when there is an eventual product launch.

Elon wants to target 20 million vehicles per year to replace 1% of the global fleet each year.

SOURCES- Tesla Investor Call, Elon Musk, Cleantechnica
Written By Brian Wang, Nextbigfuture.com

31 thoughts on “Tesla Should Have Early Access Release of Feature Complete Full Self-Driving by the End of 2019”

  1. It seems to me that the solution to Full Self Driving is to add a transponder to the license plate on every vehicle. That way you can likely eliminate most two-vehicle collisions, and if you add a communication protocol between vehicle computers, 100% of them. The AIs could advise each other of their intended course and speed, and also alert each other to road hazards, etc.

    Imagine if there was debris on the road ahead, maybe in a snowstorm at night, and your car was notified of its precise location while you were still kilometers away.

    Even collisions with other human-driven cars would be more easily avoided if their exact position and speed was known, updated every millisecond.

    As a bonus you could sell transponders to pedestrians and cyclists (dogs, cats, turtles, etc.), to help them not get run over.

    Reply
  2. Better Place was the company

    https://en.wikipedia.org/wiki/Better_Place_(company)

    There are dozens of startup article s covering the rise and fall. It sounds like they were sorta close, but the political realities of having to roll out a rather large network of expensive swap stations, the compatible EV’s, and the battery fleet to make the system minimum viable were a big stumbling block.

    Reply
  3. It sounds like the problem in the early days of railways to get a standard rail spacing. Pick (almost any) one & get everyone to stick to it has great advantages, it is just the politics that is difficult. See also getting the US to finally go metric, or deciding on a standard for almost anything.

    Reply
  4. this is silly to argue with someone about freedom.. its a personal thing and if the person feels constrained then thats their reality.. its a fair and valid concern for many people… to discredit that concern does no one any good

    Reply
  5. Car companies got cold feet over that with the flameout of the startup trying to do that in Israel. Even Tesla, with that stunt swapping event, basically stopped work. The biggest issue is that for cars, it works best to have the pack in the frame floor, but getting companies to agree to a common pack that can be removed quickly and safely is like making political sausage. The pack size doesn’t help either. There’s also the existential risk that a swappable pack skateboard frame EV would effectively disiintermediate the skateboard frame makers, leaving the body makers as the only ones with a differentialable brand.

    Now, those EV mopeds with the swappable canister battery packs in Taiwan and China appear to be starting to gain traction in their respective markets. The scale of those packs, in terms of cost and weight and charge speed, makes for a decent market fit.

    Reply
  6. You obviously don’t own a plug-in vehicle because nothing could be easier than just pluging it in at night unplugging it in the morning. This is also a lot cheaper too than using a public charging station. I don’t buy the freedom arguement either because there is no hassle of finding a station or having to waste time charging at the last minute before you go some place.

    I agree that for full adoption of EVs across all sectors of society many more public charging stations will need to be added. But only the ones outside of the cites and on the major highways need be rapid chargers.

    Reply
  7. For all those occasions, when you want to go farther than 180 km, or you want to travel more than 360 km over the day. Then you obviously need to charge on the road.
    And charging at home is an extremely HUGE PAIN in the ass!! It diminishes your sense of freedom an order of magnitude. I won’t deal with plugging my car every day. That would be a nightmare. 
    The scarcity of electric charging stations is another HUGE pain in the ass. It diminishes your sense of freedom another order of magnitude. I won’t deal with searching stations, or planning and rerouting my path to be able to get to one.
    When they can solve automatic charging at home (for example with that tesla “snake”), and they will have at least half megawatt charging stations EVERYWHERE, and the ordinary EV will have 600 km range, that’s when I will change to electric.
    In ten years the situation might be much more acceptable.

    Reply
  8. I would accept an autonomous car that can drive as well as a novice driver or senior citizen. I think the convenience factor outweighs the concerns about safety for me. That shouldn’t be the end goal though, and auto manufacturers are going to have plenty of incentive to make their self driving features better than that. Sooner or later, the liability for auto accidents is going to be on the manufacturer and the manufacturers are going to want their systems performing at least in the 90th percentile of drivers by the time they’re held responsible for it.

    Reply
  9. I think we are talking about slightly different things.

    You are, I suspect, saying that there is no rational reason to have the moral authority to demand that a robot is better than a teenager who got their licence a week ago, and who has been awake for 24 hours with a bad flu, and is listening to loud distracting radio while arguing with their friends in the back seat.

    I am saying that the voting, and political trouble making, public will be upset and will push back hard, unless the robocars are as good as the average person in good health, 20 years driving experience, and a near perfect driving record.

    Both these positions can be true at the same time.

    Reply
  10. I’ll agree that the biggest own goal Tesla has kicked is when they

    • . Officially named their intelligent cruise control system “autopilot”
    • . Splashed publicity everywhere about their self-driving cars
    • . Got all surprised when people turned the cruise control on and started watching a video or something, trusting the autopilot to be able to pilot automatically.
    • . Until the car encountered a situation that the software misinterpreted.

    That was just stupid. You don’t name a consumer product something that will kill people if they believe it.

    Reply
  11. Why not design the batteries to swap out for freshly charged ones? If you are running a fleet of vehicles that should be fairly easy. That way you can charge the battery at a modest rate whenever other power demand is low & be performing a service to the power grid so you get the power cheaper.

    Reply
  12. And I say, not until they match the worst class of human we permit to drive. Once they’re that good, what basis do we have for saying “no” to them, and still allowing senior citizens and 16 year olds to drive?

    We have, by who we license, set the threshold for safety. I see no good reason to subject computers to a double standard.

    Reply
  13. So power companies would need a substantial systemic infrastructure upgrade, and unfortunately that may require two whole decades. Though, I would make terms with a megawatt-capable charging system :).

    Reply
  14. What really annoys me about Elon is he does things like come up with his own levels of self driving (ignoring the SEA definitions) and then declares them safe without explaining what they mean.

    Reply
  15. the rational threshold for approving them: When they’re at least as safe as the most dangerous class of drivers we permit on the road.

    No, that doesn’t pass my criteria.

    We will be removing some humans from the mix, and adding robots. The humans we remove will be rich and upper middle class (because new electric cars are expensive), technically minded people who are probably interested in cars and car technology and so will NOT be (on average) the worst class of drivers on road.

    We would be removing human drivers who aren’t the worst, and replacing with robots equivalent to the worst. A net decrease in safety.

    And being realistic, the new approach always has to be significantly BETTER than the old approach, that’s just human nature in response to change.

    I’d say you can’t begin until at least robots match the average human. And probably you need something like the 80th or 90th percentile.

    Reply
  16. So your argument against self driving vehicles is that automation specialists are unfashionable “geeks” and don’t conform to your expectations of social behaviour “children”.

    Not very convincing.

    No wonder they rejected you from engineering school.

    Reply
  17. The folks working with Uber on eVTOL stuff are looking at water cooled megawatt class charger cables, so they are already roughly there. The bigger issue is the buffering power supply at the charging station to feed such a beast. You don’t take a megawatt off the pole willy-nilly, the power company will hate you for that. Tesla Powerwalls, like those that seem to now be in the process of being deployed in California blackout area superchargers, is a first step.

    Reply
  18. Is there any chance, that we will have 2500 kW chargers in ten years? That would need a 26% YOY progress for the next ten years. 250 kW is roughly ten times slower now, than refueling with gas.

    Reply
  19. What is sight and how exactly does a computer understand what is important from the literally millions of choices made while driving. Our brain is the product of millions and millions of years…we are genetically programmed to understand some things are dangerous. What does a computer focus on when it “looks” at the road?

    And yes driving is still dangerous while using the most sophisticated computer that we know of, our brain…but hey let’s throw a machine out there that has been worked on for what 50 years? That will work better…right.

    What we will start seeing as these ridiculous machines made by children start running on the roads are accidents that no one will understand. Even with our ridiculously powerful brain we will still be surprised by the antics of machines. And they will kill people…probably a lot of people.

    Reply
  20. I tend to agree: Self driving cars don’t have to be perfect. They just have to be as good or better than a reasonably large set of drivers who might adopt them. Senior citizens, for instance.

    They can improve from there, but that’s the rational threshold for approving them: When they’re at least as safe as the most dangerous class of drivers we permit on the road.

    Reply
  21. Average number of car accidents in the .U.S. every year is 6 million. More than 90 people die in car accidents everyday. 3 million people in the U.S. are injured every year in car accidents. Around 2 million drivers in car accidents experience permanent injuries every year.

    But yeah fuck it as far as you’re concerned geeks are arrogant – well people that accept the status quo as it is, like you, are utter idiots. Try to understand this is the future and it will save millions of lives and injuries. However difficult and relatively slow the transition.

    Reply
  22. Hmmm…and so the arrogance of the geeks will be on display. Sight is essential for driving…cars cannot see. What fun we will have when a fully loaded semi goes crashing into a building.

    Reply

Leave a Comment