Your speaking of what we called back in the day "the banana curve"... which was lingo for the trade off between area vs. speed (when power and manufacturability were much less of a concern). Either way you are optimizing the chip - your goals can vary. Most often people wanted a particular clock speed with the smallest area. when I think of a manufacturing plant, you want the fastest output of cars first, and then when you realize the bottleneck, the rest is to optimize floor space. Moore's law is relevant as when chips could get denser, automation is what was needed (my former industry). In the early days all this was done by hand. Then when things got dense enough, drawing it by hand (or symbolically in a computer program) was time consuming. And as things got bigger, computer algorithms were developed to automate making things faster and smaller... and then power sensitive etc... So as Moore's law occurred, these algorithms changed along with it. I'm thinking just as there was an inflection point in the 80s to develop this automation, perhaps now there is enough skilled robotic technology on the line to do some similar things now on the factory floor.
No matter what he meant, I couldn't help but think of this: Clanking replicator - Wikipedia, the free encyclopedia Self-replicating machine - Wikipedia, the free encyclopedia Replicator (Stargate) - Wikipedia, the free encyclopedia Von Neumann universal constructor - Wikipedia, the free encyclopedia The 3D printers that print themselves: how RepRap will change the world RepRap - RepRapWiki
Good stuff. But then I couldn't help but think of this: "Men have become the tools of their tools." - Henry David Thoreau
Although I agree with you to some extent the production ramping problems with the Model X had more to do with the complexity of the car then the assembly line itself. I do expect Tesla will introduce new techniques in the 3 line as they continue to improve automation.
With the Fremont factory floor space still 75% not being used, and lots of cubic space above the assembly lines not being used, I think Elon was describing his vision of how he'll set it up with more efficient programmed robots that can do more than a single task (as some on the floor already do.) I think the final assembly line of robotics and humans will accomplish more "assembly" within the same sized space of the current assembly line -- with the goal of getting the finished Model 3 traveling out the door faster than the pace of a tortoise. Example: while a robot is installing the windshield, another robot will be positioned at that same station that can also simultaneously install an item that was planned for install at a later station. Two or three robots working in the space that was previously dedicated to just one robot. Side by side or mounted on rails above - and more use of the upper space similar to how the vehicles move to the paint shop on rails high above the floor. Same production quantity or more in half the space -- allowing twice as many assembly robots in that factory as originally estimated. I think that's how and why Elon mentioned he believes his Fremont Factory will be capable of 1,000,000 vehicles per year (vs Toyota's/GM's 500,000 per year -- without Elon's robots) Elon even mentioned last year he has a crazy goal to install a roller coaster for employees to travel inside the factory. With more efficient ways of moving employees and assembly lines, the tortoise Elon mentioned would get very tired trying to keep up with completed Model 3's exiting the factory at a rate of 500,000 by the end of 2018, then a million per year just a few years later. Knowing what Musk accomplished with EV's and Rockets, I believe his production goals are very doable by concentrating on.......... "building the machine that builds the machine."
'Machines building machines'... brings to mind the way 3D printers that build the replacement parts for themselves. It also brings to mind adaptive technology where actions in a system change in response to observed needs/trends. I am imagining a factory where the robots build the robots that make the factory more efficient based on adapting to the needs of the line. Crazy stuff.
Well Elon better get going designing these machines that build the cars. In order to meet the self-imposed deadline those machines have to be designed and ordered in the next few months or so.
Who's to say the process hasn't already started. Just because he talked about it yesterday doesn't mean the idea was just hatched. He has been observing the MS production line for several years now and knew back then what the long term vision was and what might be needed to fulfill it. I think he is already way ahead of us. We just choose to live in the now.
Somewhat related, there's a push in the space industry to figure out how to manufacture on orbit. Hucking a bunch of raw material up there and then making/printing the component on orbit has massive implications. From a similar angle and slightly more on topic, if you can 'make' (whatever that exactly means at the time) a part of a car right there on the line, right when you need it, there are similarly massive implications.
If you go on the factory tour you will understand the statement, see some of it in action and after seeing how many people are involved, and some of the quality issues they are having with the Model X, it will be imminently clear.
Good to know, I really want the ramp to be smooth and fast. Delays were part of the reason I have a GL and not a Model X. I don't want our M3 reservation to end at the MB dealer.