OK, this could fall under physics and chemistry I guess. Take the human body for starters. Pay attention to where you shift your weight when you walk, how your legs move, how joints work, and so on. Scientists have been having a tough time replicating bipedal motion in a robot, and there is a lot of balance problems. So lets take a look at another way to move a robot. Pilot it with hand and foot controls? You still have a balance issue, nor would you have fine control over joint movement as you would in your own body. Pilot it using your own body as a model!

This is basically how it will work when one of the nice members here donate a large sum of money to my "build a big walking robot" fund. Wear a nylon (or spandex) suit. Two ways you can handle this at this point. Snag a crapload of laser diodes from laser pens, or buy them in bulk. Have them spaced all over your suit. Sections of your suit will be attached to the inside of the robot. You will be riding around inside this thing, with your feet on rollers. Inside your "cockpit" you will be surrounded by photovoltaic cells. When you move, the laser diodes are of course going to be pointing somewhere else. I still have my electrical engineering book, and it shouldn't be TOO hard to build a simple IC that can look at how far apart cells are activated to figure out how much acceleration and speed your limbs are generating. Another way to do this is to build accelerometers and speedometers directly on the suit. You will be (mostly) suspended inside this thing so you will have a decent range of motion, enough to walk naturally on the rollers.

Now how is this useful? Using a variable voltage output from the processor, power is delivered (relay driven) to the electrical motors. You'll have "muscle groups" that are matched to the muscular structure of the human body. Get rid of the head and hands though, not necessary for walking. When you move a muscle group, say lift a leg, the robot will activate the same muscle group at the same time, with a corresponding amount of force. So what is a robot muscle group? 4 electric winches each with a steel cable attached. They contract only. Kill power and they "relax" to starting position. Install shocks in the lower legs to dampen walking pressure.

Power source! Old 1980's 4-cylinder truck motor that operates a series of alternators, and a fat battery pack. Put all the power stuff in the groin area for a lower center of gravity, and while you're at it make the feet BIG. Kind of like a midget softball player, only 30 feet tall. If YOU lose balance easily, so will the robot. Just don't do your simulated walk too fast, otherwise the steel cable driven muscles won't be able to keep up.

Well I guess this is more physics than chemistry after all. What would be a good battery for something like this? Think hollow steel tube would be good for a superstructure (cheap)? Any idea how much tiny accelerometers and speedometers cost? The rest I can plan for. Any ideas on motors to use?

What about those new polymers that contract when voltage is applied? Do they come in sheets? How do I make them? AND! ... just what is the chemistry behind those things anyway? They're pretty snazzy.

I remember when we were discussing a mech before, but the first challenge to tackle is to get the stupid thing to walk without falling down. Most ideas about building these involve a control panel of some kind, whereas I just want to link movement to the human body. Since a robot has no inner ear to detect imbalance, if it starts to tip over, the human inside (me) will realize this and hopefully shift weight around. After that, the robot will follow human movements and hopefully be able to right itself.

Best thing is, a lot of this can be built with parts from a junkyard. Take motors from old washers and dryers, just use enough to split up the load. Get the superstructure parts from any scrap metal place, just need metal rods of the same metal. Get all the IC's for the control computer from radio shack, along with wires. Grab an old honda/toyota truck motor from a trashed truck in the junkyard, along with a couple random sized alternators. Grab transformers from any miscellaneous electrical equipment in a junkyard, make sure they match the motors you use! Get a welder from my neighbor, along with tools (he has everything). The hard/expensive part is getting accelerometers small enough, and with the right kind of output. You could just use many speedometers all over your body, and measure in microsecond increments. If you get 2 differents speeds over a span of time, bam: acceleration.

Basically, all the weight is going to be supported by the "muscles". The bone structure/superstructure just has all the motors bolted to it, along with the person. Well, you don't bolt the person to it, just his jacket. All the joints are going to be free-bending, just use ball-joints from an early 1970's ford truck.

If I have time next week, I'm going to put together some sketches of this thing then post them on here. Gonna be busy though, I have to go do some machine gun training for a while  

Humans walk by falling. If you've ever noticed, as you walk the scenery giggles up and down because you fall and then catch yourself with the other foot. We do it so smottly it's hard to realize, infants have a hard time getting a hang of it.

The problem is reproducing what the cameras pick up behind you. Sure you can get a nice accurate picture behind you, with modern digital video technology, but how to accurately reproduce it? And at what resolution? I really like this idea, but I don't think the technology is there to cram enough LED's, or whatever, in front of a person to actually make a convincing picture. Key word here is convincing.

About the spider idea, I've already hashed that one out pretty good with some of my army buddies, and sure it would work a lot better. I want a proof of concept that a bipedal version would work. Mitch, you're right about the walk by falling. The trick is to be able to catch yourself and move forward instead of into the ground. I think thats why researchers have such a problem with getting autonomous robots to walk. I reason this will be a lot easier, because the robot is just copying what you do, movement by movement. I wonder what kind of delay would be acceptable in the signal from you to the actuators? Maybe a millisecond or two? If you jerk around too much, the robot muscles won't have enough reaction time to keep up with you.

A plasma screen wouldn't work, just tip a plasma tv upside down and watch what happens! beware, this will void your warranty. You could take the same technology that the newer LCD TV's use, but do you know how freaking expensive they are? Its ridiculous! And to cover an entire mech... you need a sponsor.

I've been doing more planning on this robot, and I ran into a roadblock. I want to use steel cables for muscle groups, but winches that can pull these cables don't pull fast enough, and they are too heavy. What should I use instead? Custom electrical motors? If I use something lighter weight, it probably won't handle abuse nearly as well, and we all know how we abuse our muscles as humans...

I wish I had the money to get NEAR a plasma TV Right now I have to save up for chemical supplies and labware gets ghettoed-up out of whatever is suitable, hollowed out fluorescent strip lights as reflux condensers, RB flasks made out of lightbulbs, and when I am feeling particularly rich, pyrex cookware

I wasn't thinking of a mech, more of a stealth armor suit, predator style

you remind me of early chemists who did their own study in private, not like modern days with research institutions everywhere..

Next pic.

The stuff in the jam jar is some sodium chlorate I had crystallising nicely

I think I will add a piccy of just one of my experiments just for laughs, this one was seperating out 4-methylindole AKA skatole out from egg white using KOH.

The brown layer at the very bottom of the test tube is the 4-methylindole.