so I'm thinking pneumatics are best for exosuit... now I need execution

Ok, so... I know this is just a touch ambitious... but ultimately I'm trying to build a relatively simple exosuit, the main code of which would reside in Arduino. It would only have essential joints like knees, hips, back, shoulder elbow, and maybe some others, and I figured piston movement would be best to power them (please correct me if I'm wrong on that)

I did some research, and I think hydroulic might be best since it is extremely strong, can be fast, and I think it can be relatively accurate. The only question I have about that is where I would put the water source and how to use it... a link to an article on this would be really helpful.

Originally, I was going to go with pneumatics, but apparently accuracy is difficult with those and so is mobile air compression. Otherwise they would be great though, so if anyone has something on that...

Another option I found is motorized linear actuators, but I'm kind of worried that I would have to pick 2 between strong, fast, and accurate. If anyone has a way around this, that would probably be the best thing to use because it's much simpler than hydroulics and only requires a battery (and some fairly simple circuitry to get it hooked up to Arduino) to operate.

It'd be really nice if I can find something that doesn't cost much, but if not then until I get a job I'm mostly just looking at proof of concept anyway :P Thanks in advance for your help!

Hydraulics normally (perversely) use oil, not water.

I was going to go with pneumatics, but apparently accuracy is difficult

Pneumatics suffers from an elastic medium, but this also may provide cushioning.

I've seen rolling diaphragm pneumatic muscles capable of pretty precise control

AWOL: Hydraulics normally (perversely) use oil, not water.

Oh ok, I didn't know that but it does make more sense.

AWOL: Pneumatics suffers from an elastic medium, but this also may provide cushioning.

Yeah that's one reason why I was going to use pneumatics: so that there would be a bit of a cushion and it would be smoother, especially when idle.

AWOL: I've seen rolling diaphragm pneumatic muscles capable of pretty precise control

Thanks; I'll check that out! :)

connect a hydraulic ram to a positive displacement pump driven by a stepper motor

I’d take extreme care before putting on an hydraulic exosuit, that might break my bones on malfunction, or might hinder me from shutting it down.

Powerful hydraulic actuators tend to be heavy, the same for the pump and its power supply. Do you have more detailed plans about your mechanical design?

Hydraulics generate a tremendous amount of heat that must be dissipated, somehow. Your pump must maintain the necessary pressure for your system, whether you are using it or not. The energy applied to the fluid is stored as heat, unless the energy is used to move a motor or piston, etc.

Paul

People have been working on exoskeleton equipment for many years. You should use Google to study previous efforts. You could build up a test exoskeleton using light pieces of wood and hinges before adding motive hardware.

If at this stage you think that (the kind of) hydraulics (you'll need to use) use water then I'm thinking the timeline on this project achieving what you want will be years, if not decades.

Just saying there are quite a few aborted exoskeleton (masters/PHD) projects laying about the mech dept at the university I attend.

However! I'm not saying don't do it, you'll learn heaps, which is all good - but maybe your expectations could be tempered with a little more research?

I'd opt for pneumatics myself (and an appropriate change of scope).

Thanks again for the feedback, guys! I do agree that hydraulics would be strongest but maybe too strong to be safe, and I also didn't know about the heat generation (obviously I'm not very familiar with hydraulics).

The other day I actually talked to a friend of mine who is involved in an exosuit project at his school, and they went with pneumatics. He said it is very smooth, fairly practical, and has a rather impressive max lifting load of 400 lbs.

So basically, pneumatics appear to be the consensus here; I think that's what I'll go with as long as I can find a way to make the movements accurate and find a practical way to pressurize the air. And then obviously I'll need a way for the suit to know when and by how much to move each piston... and build the actual thing... Did I miss anything?

I'll bet the power source for the pneumatics was a SCUBA tank or similar!

Web search 'festo' along with your other search terms.

You'll find some interesting stuff.

Be prepared for $$

ahh yes, a SCUBA tank seems like it could do the trick... and a quick Google search of Festo tells me they should have some good relatable things. Thanks so much!

Not sure why SCUBA had been introduced to the discussion.

Standard industrial pneumatics will do fine.

So basically, pneumatics appear to be the consensus here;

Bzzzzzt! Pneumatic use on an exosuit will probably be unworkable. Hard to control, energy inefficient, and probably extremely dangerous.

Agree, but perhaps for other reasons - this is always going to be hard to control.

In fact I think unfeasible given the resource the OP appears to have.

State feedback approaches ... non-linear control theory... PID aint going to cut it !

Hence, 'pneumatics, and an appropriate change of scope' ;)

I’d advise you to get “Wearable Robots: Biomechatronic Exoskeletons”. To sense the wearer’s intentions, generally either force sensors or electromyographic sensors (senses nerve impulses) are used. I think emg is generally used in prosthetic devices as a wearer with severe physical challenges may have trouble even lifting their limbs. The LT1920 data sheet gives an example of a nerve impulse amplifier. Please note that in this scheme you are making numerous low resistance contacts with your body, because of the shock hazard this should be battery only and you might use optoisolators to isolate the sensor power supply from the actuator power supply.
I know of one project that is using Flexforce sensors to indicate the wearer’s intentions. The output of these things isn’t linear, but I think the data sheet covers this. It’s certainly possible to get a force sensing exoskeleton to conform to the user’s movements well, check out videos of the Sarcos exoskeleton. But as many have said, probably not that easy.
Another thing you’ll notice about the Sarcos exoskeleton is that the user’s arms aren’t strapped into the exoskeleton, there are warnings in the literature about being too tightly coupled to the exoskeleton, the exoskeleton may transfer a significant portion of the load to the wearer and injury may result. Imagine lifting an engine block and suddenly having that load transferred to your back!
As far as a project which is on a doable scale, search Youtube for James Hobson’s exoskeleton video. This thing only augments elbow flexion/extension. I would probably start even simpler than that, only augmenting the flexion of one arm. And only lifting weights I could comfortably handle without assistance, it should be easy to observe if the exoskeleton is providing assistance. If you feel compelled to move up to heavier weights, you might want to invest in a back brace first.

Agree, perhaps start with a teleoperation project.

maybe a simple concept to start with.

There is a bit of a dichotomy between teleoperation and exoskeletons. Teleoperation can be implemented by measuring the joint angles of the operator and causing the remote robot to duplicate these angles (although a little force feedback is probably nice). The remote robot follows the user. In the case of an exoskeleton, joint angles don’t provide useful information about the users intention, for example, if my elbow joint is currently at 90 degrees, it doesn’t tell the system if I intend to keep it there or flex it or extend it. A special purpose teleoperating system could probably be made to simulate an exoskeleton, but I would think it would be more complicated than just implementing the exoskeleton. Although much safer, which is certainly a concern.

zoomkat: Bzzzzzt! Pneumatic use on an exosuit will probably be unworkable. Hard to control, energy inefficient, and probably extremely dangerous.

Really? Just this morning the friend I talked to sent me a video of one of their members operating both knee joints and one arm joint of their exosuit so far, all pneumatic. Admittedly, it was suspended in a metal cage so it didn't actually go anywhere, but this was just a saftey precaution and didn't appear to otherwise interfere with the suit. He didn't get much footage of the legs, but what I saw from the arm was very smooth. I also admit it will likely be inefficient, but at this stage I can worry about that later.

1:1: Not sure why SCUBA had been introduced to the discussion.

Standard industrial pneumatics will do fine.

I think he meant just use the tank to store the pressurized air that will power the pistons, rather than having an on-board compressor. This will slightly increase efficiency... I think...

I was thinking I could have a controllable-flow valve connected to a switch valve, with each end of the switch connected to an end of the piston, then some sort of potentiometer feedback on the real-time position of the pistons, and a pressure gauge to measure the pressure in the tank. Then the Arduino could calculate how much to open the valve and what direction to have the switch, to equalize the pressure, based on three things: the position that the piston should end up in, the current position of the piston, and the pressure available to apply to it.

And of course, I would set up a simple test with one piston before going full on with the suit, probably just having the Arduino set the position of the piston based on the position of an external potentiometer. If I can't get one of the pistons to work, there isn't much point in doing the rest of them.