Compressed air is powerful enough to cinch up the lug-nuts on your car tires to the point where you can't remove them without a pry bar. That's with a machined metal mechanism, though, not a constricted air bladder.

I suspect that's what the mesh containment around the bladder is for. It gives a finite stop to the expansion of the bladder, and allows more PSI to be pumped into the air muscle without further expanding the muscle. That effectively tightens the grip without overexpanding the muscle.

If you read the initial news article, that pliant grasp is  being touted as a feature. It allows for the gentle touch needed to grasp an egg without crushing it, and then to pick up a heavy object with the same hand. The fingers are in the same grasping location. It is the variable air pressure that lets the grip be soft or firm, as needed by the situation.

ScottE -- Member (always) & Moderator (when needed)

One of the attractive features of air muscles is that they closely parallel the function of natural muscle. They produce linear tension, and bunch up when "flexed". They have stretch, or give, when lightly flexed, but are able to lock up more tightly when clenched hard.

 Depending upon the specific application, the use of compressed air for motive force can be preferrable to the use of electric servos or motors. The air muscles won't produce nearly as much RF motor noise as our more familiar electric motors do. That can mean less shielding for the electronics that might be susceptable to RF noise.

 The air muscles are reported to be much quieter in operation than an electric servo. I suppose that depends on the engineering of your air supply and vent tubes.  A pneumatic snake would seem more natural as it hissed across the floor.

You could conceivably have a large air tank in the robot to supply its working pressure, with a compressor that cycled on and off to keep the tank topped off. That's pretty much how the compressor I use for my airbrush works. 

The other technology in use for the big bots is hydraulics. It's the same technology that powers backhoes digging ditches. Lots of power there.

I worked with a Cincinatti Milicron T3 in college. That hydraulic powered chunk of iron could hold a cement block and position it anyplace, and in any orientation, within it's workspace with an accuracy of 5 thousandths of an inch. That is the type of robot you typically see on manufacturing lines assembling car bodies.

ScottE -- Member (always) & Moderator (when needed)

Those are valid points regarding the weight of a tank, and noise of refilling it. My compressor is a general purpose one, not a dedicated airbrush unit. As such, it is big, heavy, and noisy. I leave it in the storage space and run a hose to the work area I'm using.

There are many types of robots. Not all of them need to march about untethered. Every lab space I've been in has compressed air supply lines.

An air powered, pool cleaning robot might be a better fit than one that uses electric servos. Yes, I realize there are likely still electronics needed to control the air muscle valves.

How about an air-bot that attaches to the hull of your boat, and scrubs it clean while it is sitting at the dock? 

A garage-bot, adept at the use of air tools to work on a car, would likely have a ready supply of compressed air, too. 

ScottE -- Member (always) & Moderator (when needed)

Here's a nice example of the air muscles at work, as found on gizmodo

http://gizmodo.com/gadgets/robots/rob...

ScottE -- Member (always) & Moderator (when needed)