Pneumatics people, would a setup like pic related work for closed-loop controlling the volume of a balloon without attaching a positional transducer to the balloon itself? From looking on the internet, I've found that the volume-pressure relationship of a balloon is many-to-one with an initial peak as it inflates, so you can't reliably estimate volume from pressure especially when starting from atmospheric.
The balloon is just a regular latex party balloon and the pumps are http://www.robotshop.com/ca/en/sfe-air-pump.html . I don't need to spit out the exact volume. Just be able to inflate then deflate to manually-adjusted setpoints, then have the control system repeat that motion on any identical balloon.
Ideally it should be able to fully inflate and deflate a balloon quite rapidly, in 1/2 a second or less. The sense chamber would large enough that it would be negligibly stressed in doing so. I also realize that compressing the balloon will cause volume to decrease without registering in the sense chamber, but this is an acceptable compromise.
>>1010060
what is even the point of this?
is this an experiment?
homework?
>>1010269
soft robutt experiment
>>1010349
Now that you're being more clear, rather than a pure soft robot, which sounds like a pain in the ass, I would make a hybrid soft/hard robot with joints and rotary encoders.
It'd be a lot easier than building a circuit that PIDs based on volume vs ambient pressure.
>>1010381
Unfortunately it can't have hard joints or bearings, kind of goes against its USP I wanted to have few or no electronics on the soft part, with the movement being controlled through an air line "tether" like in the above pic. That way it would be easier to build swappable bodies and use the same pump/controller thinghamajig for all of them.
That said I'm probably going to compromise and glue Spectra Symbol flex sensors to the thing with rubber cement or something. That might make it more robust against changes in atmo pressure etc. and simplifies the controlling part to just two pumps and a manifold, but now it's got wires and I need to make my pals glue flex sensors in the exact same spots.
>>1010381
Soft robotics are shit purely because they need to be tethered. Unless youre talking about something big enough to carry its own air tank.
>>1010403
Youre not going to get wild precision with soft robotics. So a 20kpa drop isnt going to make much of a difference. Not to mention your feed in pressure is measured as the delta ambient pressure
Vent it to atmosphere, that's what everyone does.
Hahahahaa good luck with volume control. One issue you will run into is hysteresis and the viscoelasticity of rubber. The rate of stretch and unstretch will change things and rubber will slowly relax a bit under tension. We can't model viscoelastic. Oh and did I mention that elastomers set point changes with stretches? You know how you stretch a balloon and it inflates bigger?
You need position sensing m8.
>>1010403
They suck. You are better off making your own strain gauges. No seriously.
For best results do this:
http://onlinelibrary.wiley.com/doi/10.1002/adma.201506234/full
Or microchannels filled with gallium
