Too much automation in T3 final assembly?

Feed The Trees said:
https://techcrunch.com/2018/03/28/t...ation-in-model-3-final-assembly-analysts-say/

The story basically says there is a known direct decrease in quality of the finished product when you have robots do the final assembly vs humans, and it saves on average about 5 hours.
Click to expand...

At the risk of sounding biased, I have to say this analysis seemed flawed. I mean, this robots-can't-do-quality-assembly-work take seems better suited to the early nineties. I recall some automaker recently bragging about the human component of their final assembly, but most automakers are equally bullish on the capabilities of their robots. I dunno. I just find it hard to believe that robots can't be as good or better at doing quality work as humans.

Also, according to this tweet, "The paper by Gorlach and Wessel rests heavily on automation cost analysis from a 1993 paper by M. Hartmann."
 
Interesting, did not know that 1993 component. So cost wise it may not be right, but how about the issue of the robots having a known decrease in final product? I guess the theory is they can put the bolt in precisely the right spot, but if this panel is juuuuust a hair off then a human eye can make a little tweak here or there and present a better end product. I have zero manufacturing experience though but I could see that being the case.
 
Feed The Trees said:
Interesting, did not know that 1993 component. So cost wise it may not be right, but how about the issue of the robots having a known decrease in final product? I guess the theory is they can put the bolt in precisely the right spot, but if this panel is juuuuust a hair off then a human eye can make a little tweak here or there and present a better end product. I have zero manufacturing experience though but I could see that being the case.
Click to expand...

It just strikes me that a robot, properly programmed, would just do it right every time. I would love to visit the plant and ask all these questions.
 
The robot will do it right, but not every piece that comes out of the plant is exactly the same, there are tolerances. When enough are within tolerance but not perfect the fit starts to really go awry was you build to last part. This is where a robot will blindly assemble where an eye would possibly adjust.

Or maybe I'm talking out of my tailpipe.
 
Feed The Trees said:
The robot will do it right, but not every piece that comes out of the plant is exactly the same, there are tolerances. When enough are within tolerance but not perfect the fit starts to really go awry was you build to last part. This is where a robot will blindly assemble where an eye would possibly adjust.

Or maybe I'm talking out of my tailpipe.
Click to expand...

You are right, this is the argument. Tesla's answer to this is tolerance to 1/10000" for the whole car. If each part is built to this high standard, then fit issues will go away. When cars are assembled with poor fitting parts, they have to do more body work at the end of the production line to make everything look good, it is not just a matter of adjustment.
 
Feed The Trees said:
The story basically says there is a known direct decrease in quality of the finished product when you have robots do the final assembly vs humans, and it saves on average about 5 hours.
Click to expand...

At the risk of being blunt: Bullsh*t. Automation is a tool. Like any tool, it can be used well or used poorly.

A poor workman blames his tools. -- English proverb

Watch a few episodes of the documentary series "How It's Made". You'll see plenty of automation, including high-speed automation, making everyday products. Generally speaking, there is no inferior quality involved, assuming the assembly line is properly set up and fine-tuned.

There is not really any mystery here, altho there is certainly plenty of hype coming from Tesla Inc. The exact details about the various delays in ramping up Tesla Model 3 are being hidden by Tesla -- as they would be hidden by any company bigger than a mom-and-pop operation. As they say: "Does Macy's tell Gimbel's?"

The overall problem is pretty obvious: Tesla is not allowing enough time for doing the necessary testing and fine-tuning. The motivation for that is understandable: Tesla is trying to ramp up production from relatively low volume to high volume much, much faster than is the norm for the auto industry. But just because they have the motive to do so, doesn't mean that practical difficulties are going to disappear. Tesla's production is as subject to Murphy's Law as any other complex human endeavor.

Anything that can go wrong, will go wrong. -- Murphy's Law

Murphy was an optimist. -- Anonymous

The perversity of the universe tends towards a maximum. -- Finagle's Law
-
 
Last edited:
Feed The Trees said:
...how about the issue of the robots having a known decrease in final product? I guess the theory is they can put the bolt in precisely the right spot, but if this panel is juuuuust a hair off then a human eye can make a little tweak here or there and present a better end product. I have zero manufacturing experience though but I could see that being the case.
Click to expand...

Having worked in a machine shop, I can tell you that the human eye is often an inadequate measuring instrument. That's why we have measuring devices such as micrometers. If the tolerance is that important, then the automated line should use a laser scanner to ensure a hair-fine positioning.

The real limit to automation on auto assembly lines back in the 1980s*, for example GM's failed Saturn line, was that the "robots" back then were not flexible enough. They were too limited and could not easily be repositioned, repurposed, or reprogrammed to perform a different task. The Saturn manufacturing lines were ultimately abandoned because they could not be used for anything other than building the one exact model year of the car model they were set up to build. Contrariwise, many or most auto assembly lines can make more than one model, and many variants on those models. In a documentary examining the failure of Saturn, it was pointed out that in Japan, machines and automation were (at that time) being used more to assist humans on auto assembly lines, rather than vice versa. Humans are much more flexible, and can usually be "reprogrammed" with just a few hours of training for another job on the line.

*You were talking about automation in the 1990s, but I'm going to talk about what I know at least a bit about rather than what I know almost nothing about.

As Domenick says, things are different today. Robotic arms are much more flexible, much more versatile, and much more easily repurposed and/or reprogrammed, than they were just a few decades ago. Of course humans are still much more flexible and versatile than robots for nearly all tasks, but the cost of human labor generally keeps increasing, while the cost of robotic "labor" keeps going down.

Furthermore, Elon Musk is right about the speed of assembly line work. The eye-blurring speed at which many assembly line machines move on the "How It's Made" series is often amazing. There's simply no way that human workers could ever match that speed. So, the potential for much greater throughput is there, if human workers are removed from the assembly line and all replaced by robots. Musk is certainly right about that. But very clearly, putting that theory into practice is proving to be much more difficult than he anticipated. That doesn't surprise me at all. If substantially increasing throughput by using near-100% automation on an auto assembly line was that easy, then it would already be a widespread practice.
-
 
Last edited:
Roy_H said:
You are right, this is the argument. Tesla's answer to this is tolerance to 1/10000" for the whole car. If each part is built to this high standard, then fit issues will go away. When cars are assembled with poor fitting parts, they have to do more body work at the end of the production line to make everything look good, it is not just a matter of adjustment.
Click to expand...

Isn't that what their 'build fast fix later' as Reuters put it issues are about? Along with this https://www.greencarreports.com/new...arts-need-substantial-rework-after-production
 
Feed The Trees said:
Isn't that what their 'build fast fix later' as Reuters put it issues are about?
Click to expand...

Do you think this doesn't happen with other auto makers' cars, when manufacturing starts on a new model? Do you think they don't have problems show up which have to be fixed by hand after the car rolls off the assembly line?

Honestly, it gets very very tiresome seeing all the Tesla bashing that is nothing but reporting what is a perfectly standard procedure for auto makers, and trying to twist it into something unusual or "wrong" just because Tesla is doing it. There is a lot of that going on at Desperately Seeking Liars Seeking Alpha; we don't need that cabbage here!

Tesla does have some real problems. Why don't we stick to honest discussions of those, rather than all this made-up FUD bull pucky?
-
 
Last edited:
I'm just passing it along and asking questions to understand the issues. The papers seem logical, I don't follow the rest of the industry to know what's normal and what's Tesla specific.

Sent from my Nexus 5X using Tapatalk
 
This is what the bad kind of union wants. The very one that got blamed for GM's collapse when that too was nonsense. That union need to be focused on getting a high indexed GAI for its members not this nonsense. These are just PR lies.
 
Robots wont have the Monday morning or Friday arvo Blues or bad days.
I remember doing a reading test for picking up errors once and most people only picked up about 10 errors out of 100 to show QA didn't work very well , its all in the procedures
 
Feed The Trees said:
https://techcrunch.com/2018/03/28/t...ation-in-model-3-final-assembly-analysts-say/

The story basically says there is a known direct decrease in quality of the finished product when you have robots do the final assembly vs humans, and it saves on average about 5 hours.
Click to expand...

I believe this would be the right thread to drop this off:
Musk automation tweet.webp
 
Feed The Trees said:
So no I have to wonder if Elon is acknowledging a mistake or using a public scapegoat to buy himself time. Never know, crafty fox.
Click to expand...

When it comes to Elon's use of Twitter to promote his companies and his agenda, I'd say his motive(s) for any individual Tweet, or any series of Tweets, is far less important than how the public reacts to it/them. Regardless of any other motives he may or may not have for any individual Tweet, or series of Tweets, his primary reason for Tweeting is to shape or shift pubic opinion in favor of his agendas. And despite the occasional instance of foot-in-mouth, he has proven to be remarkably adept at that!
-
 
Pushmi-Pullyu said:
When it comes to Elon's use of Twitter to promote his companies and his agenda, I'd say his motive(s) for any individual Tweet, or any series of Tweets, is far less important than how the public reacts to it/them. Regardless of any other motives he may or may not have for any individual Tweet, or series of Tweets, his primary reason for Tweeting is to shape or shift pubic opinion in favor of his agendas. And despite the occasional instance of foot-in-mouth, he has proven to be remarkably adept at that!
-
Click to expand...
Yes, still curious though if it's truthful or riding the coattails of a theory, a theory that was roundly debunked by many, citing it's reliance on a very old paper.

So reasons do matter, if you're an investor. Is it the truth or is it just to buy him time?
 
Feed The Trees said:
Yes, still curious though if it's truthful or riding the coattails of a theory, a theory that was roundly debunked by many, citing it's reliance on a very old paper.

So reasons do matter, if you're an investor. Is it the truth or is it just to buy him time?
Click to expand...

Probably him being truthful, he has made similar comments in the past when he felt he set an overambitious goal that ended up causing trouble. See: Falcon Wing doors.

There are lots of tasks that humans are better/cheaper at performing. Making adjustments on the fly is probably far more simple when you just need to explain it to 5 humans. Having engineers reprogram the automation for every little adjustment probably doesn't come cheap.

Plus, a lot of the sweet tax deals that corporations get require companies meeting certain employment requirements. That may be a factor that makes humans ultimately cheaper as well.
 
If he's being truthful then that's great, but I will point out that all the Tesla fanboys around were out in droves to discredit the too much automation report. Guess it's time for them to eat some automated crow.
 
You must log in or register to reply here.
Back
Top