TMC is an independent, primarily volunteer organization that relies on ad revenue to cover its operating costs. Please consider whitelisting TMC on your ad blocker and becoming a Supporting Member. For more info: Support TMC

Future of Automation and Alien Dreadnought

Discussion in 'TSLA Investor Discussions' started by generalenthu, Jul 13, 2018.

  1. generalenthu

    generalenthu Member

    Joined:
    Jun 10, 2015
    Messages:
    655
    Location:
    Vienna, VA
    Elon has gone back and forth with the level of automation. In early Q1, they got rid of the conveyor belt system, which they were hoping to make work only a few weeks earlier. It felt like it did not impact anything else. A few days later, if I remember correctly, they said they were hopeful of other automation on the line working. Now they're saying automation is much worse. I remember them obsessing over wiring harnesses being installable with robots, at one point. They also said the ability to automate manufacturing is going to be crucial and a long-term differentiator. They were going to test and run machines at a rate higher than specified, but now it looks like there are issues with vision even in the basic scenarios. Where does this leave Tesla with respect to model 3? Can the automation substantially reduce the man hours per car to ~20-30 hours, despite doing a lot of stuff in house.. How different will the design be for model Y? This to me was the biggest takeaway from reading the interview transcript. Hopefully someone probes Elon to talk more about automation during the earnings call.

    Please share your thoughts. For reference, in the conversation with Bloomberg, This is what Elon had to say about automation and Alien Dreadnought:

    What's the long-term plan for that new assembly line inside the tent? I think the confusing thing for most people is that you now have two apparently different processes producing the same car, one with more humans and one with more automation.


    A lot of the hoped-for automation was counterproductive. It's not like we knew it would be bad, because why would we buy a ticket to hell? We don't actually want to go for hell. We just didn't realize it was a ticket to hell. We thought it would be good, but it was not good. That applies to a great deal of the automation. A whole bunch of the robots are turned off, and it was reverted to a manual station because the robots kept faulting out. When the robot faults out—like the vision system can't figure out how to put the object in—then you've got to reset the system. You've got to manually seat the components. It stops the whole production line while you sort out why the robot faults out.

    What’s next with “the machine that builds the machine.” What’s your current thinking about the “alien dreadnought” [Musk’s term for a hyper-automated factory]?

    Let me just give you a tour of the whole giant machine. It will blow your brain right out of your skull, OK? It is so crazy. There are parts of it that are completely automated, no person there at all. And then there are parts of it which are completely manual, no machines there at all. Then there are parts of it that are partly automated and partly manual.

    When you talked about automation before, it was kind of like, you know, cars are going to be moving out of the factory faster than humans can move. So you can't have humans involved in the process. Now you've got humans heavily involved.

    You can only move as fast as the slowest thing in the system. I didn't say this would be done immediately. I was just saying that is where it needs to be in the future. And there are definitely parts that move too fast for people.

    Part of the problem is that the designing heads were naive about manufacturing. Just because we have something that works great in a simulation does not mean that it works great in reality.
     
    • Like x 1
  2. generalenthu

    generalenthu Member

    Joined:
    Jun 10, 2015
    Messages:
    655
    Location:
    Vienna, VA
    After posting this, I saw @vgrinshpun make a good point on Twitter that the robots while idle for now, might only need a vision software upgrade to get them humming again. I think there's a lot of validity to it.

    Screenshot_20180714-015754_Twitter.jpg
     
  3. neroden

    neroden Model S Owner and Frustrated Tesla Fan

    Joined:
    Apr 25, 2011
    Messages:
    10,476
    Location:
    Ithaca, NY, USA
    So here's my thoughts.

    What most people don't understand about automation is that *each individual step* requires a *totally different approach* to automation. I really started understanding this from watching too many episodes of the TV show "How It's Make". It's not some sort of generic "automate things". Each step you try to automate requires a particular invention to automate *that step*, which doesn't necessarily replicate to any other step.

    This is why the automation expert hired from Lego and the one hired from Audi are not redundant. They're automating different processes; the Lego guy is probably working on small parts (particularly the tricky problem of getting them to line up in the right direction when you pour them out of the bin into the machine which installs them, something Lego is expert at) and the guy from Audi is probably working on large assemblies.

    Tesla tried to automate *everything*. Some of their attempts succeeded, resulting in actual breakthroughs: they can automate stuff nobody else can automate yet. This is an actual, significant advantage which should translate into a better cost structure for Tesla, in the long run. The fully automated seat production line seems to be a big accomplishment, as is the "golden wheel" for breaking in the suspension and aligning the car, and the automated bolt insertion (which has stymied automaters for decades). Other automation attempts failed, and they had to go back to manual methods.

    So, a lot of the hoped-for automation was counterproductive. And some of it worked. This meant they spent a lot more CapEx than was really desirable, since maybe half of it didn't work out, but they got some permanent advantages out of the half that did work out. As long as the company has enough cash to avoid bankruptcy, it'll be a long-term advantage.

    They made some good choices in the automation attempts, specifically in "designing for automation". Often the best way to automate construction of a product is to redesign it. Switching from square-ended bolts to tapered bolts makes it a lot easier for the robot to put the bolts in (this is a specific example which was reported). Replacing thousands of individual bundled wires (very old-fashioned) with ribbon cables makes it easier for the robots to thread the cables through the holes they need to go through, which is also a big improvement. I don't have it confirmed but I think the robots are still threading the cables though the holes. But the robots still can't plug the cables in successfully, so they went back to plugging the ends in by hand (this is confirmed).

    They made one really big mistake, which was packing the manufacturing equipment too tight, leaving no room for reconfiguration of the factory line. (Musk was actually talking about deliberately packing everything tight.) This was really an unforced error and probably accounts for why nobody has heard anything about GA2 -- in order to make it match GA3 they probably have to literally move everything on the entire line. I suspect GA3 corrected *this* error.
     
    • Helpful x 6
  4. vgrinshpun

    vgrinshpun Supporting Member

    Joined:
    Apr 5, 2013
    Messages:
    5,856
    Location:
    PA
    GA3 is actually consists from two parallel lines per the reports after factory tours. This is also consistent with Elon description during the annaual shareholders meeting. Tesla never used GA2 label, it was thought of in enthusiast community based on avsilable information. So GA3 consists of line 1 and 2, each capable of 2.5k/week without automatic conveyance system. Adding modified conveyance system seem to be a part of Tesla plan to ramp M3 to 10k/week in 1H 2019
     
    • Informative x 4
  5. defc0n

    defc0n Member

    Joined:
    Sep 30, 2016
    Messages:
    613
    Location:
    Indiana
    Software is the hard part. Hardware is the easy part.
     
  6. neroden

    neroden Model S Owner and Frustrated Tesla Fan

    Joined:
    Apr 25, 2011
    Messages:
    10,476
    Location:
    Ithaca, NY, USA
    Okay. Thanks for digging that out. Tesla is using very weird numbering. I guess the S/X line retroactively became "GA1", they used "GA3" for the Model 3, and they just skipped "2" entirely? Goofy. I would never have expected something so silly.

    It makes "GA4" make even less sense (it should be GA3-C) and requires distinguishing between the two GA3 lines (GA3-A and GA3-B perhaps), so it's a *really* confusing numbering system.

     

Share This Page

  • About Us

    Formed in 2006, Tesla Motors Club (TMC) was the first independent online Tesla community. Today it remains the largest and most dynamic community of Tesla enthusiasts. Learn more.
  • Do you value your experience at TMC? Consider becoming a Supporting Member of Tesla Motors Club. As a thank you for your contribution, you'll get nearly no ads in the Community and Groups sections. Additional perks are available depending on the level of contribution. Please visit the Account Upgrades page for more details.


    SUPPORT TMC