With robotics being what it is today, why aren't there giant mecha's yet? So I started thinking, can we build them right now, with current tech?
And I haven't really encountered that much trouble, but then again, what do I know.
So here's a rough outline of my current idea, feel free to poke holes in it or, preferably, suggest something better:
Based on the following, I went for an "all-electric" approach:
- hydraulics: not too bad speed/strength, but angular range of motion of a joint is terrible. Acceptable precision.
- pneumatics: very fast, can have speed and precision trouble and the range of motion is still terrible.
- mechanical drive (connecting actuators to the main engine with pushrods or pulleys or whatever) is probably the worst option.
That leaves an option which doesn't seem to bad: electric torque motors everywhere. With a large radius, and active cooling (fluid-cooling for the high power parts such as knees, hip, elbow, shoulders, air cooling elsewhere). I did some calculating, and with current tech you can easily make your mecha strong enough with torque motors. I know, it's weird. But let's see:
Let's take one from this list, just to prove the point:
http://www.etel.ch/torque_motors/TMBA diameter of 80cm is acceptable for a 5 meter tall mecha (that sticks out only 40cm at each side of the joint, and that doesn't look as bad as you may think it does), so let's take the strongest one of those: 6.4KNm continuous, 11.2KNm peak. Let's put in the most demanding place, the knee, and assume a worst case angle of 0 degrees, and assume both leg sections have a length of 1.2 meter. That means your mecha can weigh about 550kg. WTF? Not good enough! No, but no one said you had to use just one. While it has a diameter of 80cm, it has a width of only 15cm. You can easily use three (or heck, why not four), giving you a max weight of 1600kg. That's more like it, and remember: you don't do this. This worst case corresponds to squatting on one leg. Why would you do that? You actually have two legs, and you're not even going to squat to zero degrees when preparing for a jump - and when you're jumping, you can turn the dials to eleven and use peak-torque, which is nearly twice the continuous max torque. Unless my physics suck (not entirely out of the question), this will work out just fine in terms of raw strength. You should be able to jump, run, climb and throw one hell of a punch.
Then there's the issue of power supply. Carrying around a generator and a flammable liquid works alright and is the usual solution, but if you look into Aluminium-Air batteries there is something to talk about there too. Alu-Air batteries have a lower energy density (9MJ/kg according to some sources) than, say, diesel fuel (45MJ/kg), but that's not the only factor. For Alu-Air, you don't have an engine and a generator, and that space is not insignificant - you'd need a pretty sturdy engine and generator. Also, being a battery, mechanical failure is unlikely unless moderately damaged, and it works great under non-constant load, which is what you'll have in a mecha - running, jumping, punching, etc? Such great bursts would require energy storage to smooth out if you went for a combustion motor + generator (unless you used pneumatics, which I chose not to).
Slightly into future-tech: both battery tech and electric motor tech are improving as we speak. These are active area's of research. If it works well enough
now, imagine how well it will work a decade from now, or even just with the latest research models.
Now, the controls. They have been subject of much debate. Those "hands and feet only" style controls (joysticks and pedals, seriously?) don't offer nearly enough degrees of freedom - the entire idea is to have a tactical unit that can do whatever the heck, not just a "tank with legs". I think the best option here is keeping most of the body immobile and sensing the muscles of the pilot. That also lets you give the mecha more flexible joints than the pilot has - it senses which way you want to move a joint at what strength, not where you want to have it, so it could easily "go further" than a human joint. That's also no problem at all for a torque motor. It shouldn't be too hard to train for that. Taking such direct control also means you don't need very complex "walking algorithms" and systems to keep balance - the pilot should be able to do it himself, using his natural instincts, feeling for balance, and some training.
We're not quire there yet with reading input directly from the brain, but we're eerily close. That's the sort of tech no one knows exists - it does, but it hasn't, to my knowledge, been miniaturized yet, and the response times are bad.
That works great for legs and arms, but not so much for hands. There, you'll want tactile feedback, and that sort of tech already exists as well. Also, you'll need some way to control "other stuff" (turn night vision on and off, climate control, weapon systems, etc). So I propose that your left hand (or right, if you're a leftie) doesn't control a robotic hand but instead manipulates a touchpad. That makes the left arm of the mecha a good place for fixed weapons (no lasers yet, but a nice gatling gun should work), which may seem odd at first, but you'll get used to it, and you could still hold a weapon in your right hand. I'm not 100% satisfied with this approach though, so if you have a better idea, that would be great.
Sensors (vision etc): you'll want this to be good. I'm thinking stereoscopic projection on the eyes, with extra info overlaid (also the virtual buttons and stuff that your left hand manipulates, if you can't see them you can't find them), looks around when you turn your head. And you know the drill: can zoom well, can draw boxes around targets, has infrared mode, and generally tries to provide useful information without getting in the way. That already exists, too.
So, thoughts? Is this even on the right track?
