[GlenTheMachine]

There are a lot of these. You see a couple of papers on different types of linear actuator materials every year at ICRA, just to pick a robotics conference at random.

To date they’ve all had the same problem, which is that their energy efficiency is usually less than 5%, and often less than 1%. Compare this to a decent brushless DC motor, which will have an energy efficiency of 85%. Often their bandwidth is very low as well, like less than half a hertz. So they work okay if all you’re interested in is pull force and compactness, but they turn out to be really unsuitable for most robotic applications.

Note that this particular article gives neither an energy efficiency nor a bandwidth figure.

[clairity]

yup, as part of a research project way back in college, we even made crude artificial muscles out of mesh wire sheathing[0] and custom-formed long, thin rubber "balloons" actuated using compressed air. the efficiency was horrible, but we were primarily interested in the gait dynamics of our robot, not efficiency. it was a fast little bugger though!

[0] something like this: https://www.donathenrc.com/mesh-sheathing

[ls612]

What kind of efficiency do human muscles get?

[GlenTheMachine]

Depends on the human and, I suspect, the task. Also, it's hard to compare since muscles are not powered electrically, they are powered chemically. But the estimates I'm familiar with say somewhere north of 50%. Wikipedia says 25%.

Human muscle also has a bandwidth in the ~2-5 Hz range.

See:

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3144848/