Free PCs Not AfFordable
rakerman writes: "Ford Cancels Computer Giveaway Program. I guess their 'Model E' program turned out to be an Edsel." We did at least one story about this at the time (and a Katz essay). A lot of people pointed out that the United Auto Workers union was the driving force (ha-ha) behind this program initially.
Spoken like someone who has never set foot in an auto plant.
I work for Saturn as a Controls Engineer, and you can forget about getting rid of (many) more folks through automation. The low hanging fruit has been plucked, that being the welding robots, and in our case, some of the powertrain operations.
The simple truth is that humans are still required to do most of assembly (urethaning the glass & roof not included), and to get the fit and finish of a car looking good. It will be some time before these operations are relegated to robots.
And why bother? Since if breaking the union is what you want to do, then the way to do that is to set your plant up as a "X" plant. That is where your plant does nothing but assembly (as opposed to doing the stamping, welding, painting, etc). The way this works is that you bring in the guys who make the radios, and have them install them on the line instead of a UAW worker. This not only cuts your overhead, so it's cheaper (it's pretty much certain that these guys aren't going to get UAW scale -- and never mind the benifits), but you also lessen the impact of the union.
It's already happening in South America, Ford does it, I think that we do it too (that being GM). Volkwagen planned on doing it too (in fact they stole the plans from us, and lost in court over it)
You can buy a nice Staublii or Motoman robot for around $80,000.
Of course by the time you generate controls drawings, mechanical drawings (for end effectors and fixturing) and develope the application you are problably talking closer to $120-$150k for one robot.
We very seldom use what most here would concider a computer for machine control. Besides that, I have working on a project where we had to handle 33 variations of the same part on one piece of assembly and test equipment. The entire project was less than $2 million. The cycle time for a completed part was 10 seconds.
An Allen-Bradley SLC 5/04 processor with 16k (yes, that is k) is typical of what we use for machine control. I know this seems a small amount of RAM, but when you are primarily dealing with bits not much is required.
I have to disagree with you here. In some cases, the computer can make the dicisions much more quickly and reliably than any human on earth. One example I have personal experience with is machine vision systems. I did a project that involved inspecting disk pack assemblies that were assembled by hand and then orbital formed them in place. The operator would swear that he assembled them correctly when our machine rejected an assembly. Upon inspection, it was proven every time that the operator had in fact assembled them incorrectly. During final exceptance we even rejected an assembly because an eyelash was in the disk assembly. Our customers were very impressed and it also made a strong case for doing this assembly in a clean room.
Actually that is $364k in 35 years, but you left out things like benefits, social security taxes, and unemployment insurance that the company has to pay on top of the emplyee's wages.
OK, I could go on for quite a while but I am done for now...
I know what the Internet is, what the hell is this Interweb business?!