FAUB robot program failure

An airplane is a giant tin can. Unlike a beer can, it’s big enough it has to be riveted together. What is a rivet you might ask? It is a permanent fastener with a thicker piece on one end. You drill a hole on the thing to be fastened, stick the rivet in it, then you hammer on the other end to make the other end thick also, which holds the bits of thin stuff together. There’s variations of this basic idea with tubes, pop rivets and so on, but that’s the basic idea. The ones which are used on airplanes are called “solid shank rivets“ and work like the ones you’d see the three stooges install in a girder, where it has a big head and a solid shank (made of aircraft aluminum), and someone hammers another head on it on the other side of the aluminum. So, you need a guy to hold the rivet with a bucking bar, and a guy to hammer a bead into it with what amounts to a pneumatic jackhammer. They need to work together on opposite walls of the aluminum; so, inside and outside the airplane, just like Moe and Curly on opposite sides of the girder.

Youtube doesn’t allow me to poast the video; jerks

This involves considerable craftsmanship and has more or less been done the same way on aircraft since they started building them out of aluminum. You drill the hole, then stick the rivet in, and Moe and Curly hammer it down as a team. Then they do another one. There are various tricks to go a little faster, like drilling most of the holes in advance on some kind of semi-automatic rig, but essentially the technique used to hold your airplane together is a form of human craftsmanship, literally meticulously done by hand, which dates back to pre-historical times. An ancient Roman engineer would know exactly what was going on in the Boeing factory, as most Roman military equipment used rivets. Hell a bronze age barbarian familiar with metal working would recognize this technique. How it looks:

MBAs hate shit like this, of course. Drives up cost and provides nice jobs for armies of skilled laborers. You could use that money to hire more MBAs with computers! Why waste money on working class Seattle area rednecks? So, some HBS genius in post McDonnell management era Boeing dreamed up the idea of using robots to do this process. If you’re an MBA this would seem like an obvious thing to do. After all, we live in the 21st century, and those Optimus and Boston Dynamics videos look really cool. Enter the Fuselage Automatic Upright Build (FAUB) robots.

Of course, if you’re an engineer, or you read my blog you know that all of these videos are very very fake and totally gay, robots are mostly retarded, and FAUB is likely to have problems. Superficially it seems like a robot should be able to coordinate putting in a rivet with another robot on the other side of the fuselage. The MBA might even reason that they use robots to assemble cars with welding and so on. But, those kind of robots aren’t trying to solve any open problems in robotics. If you go through my list, you’ll see that something like this would involve solving at least a couple of these open problems, which, BTW, remain open problems. I mean, if you’re trying to ship something that works, you can and should cheat as much as possible to get to a working robotic riveting pair, but apparently they weren’t able to cheat enough to get by. Probably works on flat panels, which is orders of magnitude easier for reasons that should be obvious.

The contractor was plausible enough. They sell real automotive robots that work. They even had the good graces to remove “aerospace” from the things they claim they can do now that they’ve failed to deliver. On the other hand, they still list e-commerce and house-building as things they can do, and I have grave doubts that they ever actually have done these things either (maybe e-commerce). I guess it’s good to have aspirations. Here’s the dumb thing in action, while Boeing still has the stones to keep their failson on the internet (the nose-ring woman engineer is a nice touch):

Like, you can immediately see why this monstrosity isn’t going to work, though nose-ring girl boss prominently featured is a pretty good indication in itself. It’s huge, obviously expensive, ridiculously slow and obviously not very accurate on curved surfaces (probably any kind of surface for this retard thing), and is ministered to be a bunch of working class people and expensive looking engineer types, all of whom would be more productive hammering in rivets rather than fucking around with this ridiculous contraption. After a few years of flushing what looks like a few hundreds million bucks down the toilet, Boeing laid off the robot and used the savings to invest in assassinating whistleblowers.

Supposedly Airbus has actually managed to accomplish this, by contracting with Spanish robotics companies instead of German ones (current year Spanish engineers are better than German ones; sorry German bros: no comparison, ancestors cry). You can see their approach is more likely to succeed as it takes more from CNC and doesn’t look like it’s going to fall over. I still doubt it saves them any money doing it that way. Also note the conspicuous lack of nose-ring girl bosses.

This hasn’t fully played out yet either; the Boeing robot had a lot of nice press for a while, and in this video anyway, it’s not working inside an airframe, but it’s interesting to contemplate why they might succeed where Boeing failed. I make fun of the nosering engineer lady, but it’s an important tell from the McDonnell era. They were really interested in looking good and modern and having progressive engineer lady to look at rather than the robots themselves. The Airbus video just shows the robot doing its thing. There is also the stability issue. the Airbus robot uses ball screws like in a high quality CNC machine tool. This makes motion extremely reproducible. No such thing with Boeing’s robots; I think they use standard strainwave joints. Those are pretty good joints too, but nowhere near the precision of ball screws, especially the way the thing is wobbling around. Mostly this is a game of angles: aircraft fuselages curve, and you need to be normal to the surface at both sides; big hulking thing with few joints is better at this than skinny shaking thing with too many joints. Let that be a lesson to those who believe marketing happy talk about alleged progress in humanoid robots; riveting airplanes together should be a piece of piss compared to lots of other things humans do routinely (like folding clothes).