Back! With more Weirdness!

[vacpress]

Hey Everyone!

I know.. It has been months! But, I return with a little mechatronics 202 homework problem I made up for myself as I begin to get into haptics and haptic research (See Wikipedia).

Anyhow. Attached and linked is a 'worksheet' I have made to help me figure out how to tackle this rudimentary 1-axis force feedback interactive machine.

Essentially I wanted to make a tug of war machine, but after some tinkering with gears, belts, and motors, I decided to just go super simple, and do a 'Stick'.

If any of you servo geniuses out there have any advice about ways to attack this, let me know.. I have considered resistive pressure sensors or similar as well.. But I do not know.. I also considered various scenarios with springs and linear encoders.
<img src="www.robertguyser.com/1axishaptic.gif">
Thanks!

[StealthDumpKits]

why not try something viscoelastic? I'm reminded of hydraulic counterbalance systems. Or has this been done with vehicular active suspension?

[vacpress]

why not try something viscoelastic? I'm reminded of hydraulic counterbalance systems. Or has this been done with vehicular active suspension?

i am not sure what viscoelastic is... Anyone with a servo CNC system have any experience with what it does when it hits an obstacle?

If it were to have its drive current cut down to a safe level in proportion to how hard the obstacle is, could it be setup to stall out selectively at different 'pressures'?

Thanks..

[vacpress]

i found this idea in a patent search.. even though it was meant for art project\research, i have changed it slightly.

thanks for the help.

[aspenelm]

Hello, I saw this paragraph in the manual for some servo drives I sold a while back. Don't know if this helps but kind of sounds like what you want to do...

On the SSt-1500-AXX (advanced positioning) setting the Mode line
low will put the drive into a torque/force limit mode that will limit
torque/force in the positive direction, negative direction or both. The
fold-back torque/force limit and direction is set using SSt-QuickSet’s
“Inputs and Limits” window. This mode is useful for clamping objects or
allowing the axis to become compliant.

[vacpress]

aspenelm

thanks for the information! this is actually an implementation of just about what i was talking about, yeah.. they had it built in as a feature, apparently.. it was my entire application, in a way.. i wonder how they did it!

if i ever thionk of this project again, i will do a patent search on their product... i wanted to make an art installation.. so, i dont think it would be an IP issue!

thanks.

[unterhaus]

I don't know how far you can get with monitoring current, but that's one way of doing it. An unloaded motor is going to be using significantly less current than a loaded motor, and small variations should be fairly easy to measure. A small resistor in the one of the windings should get you the current.

[daedalus]

Hi,

Are you building the hardware for this project? it looks very similar to the needle insertion simulator done by quanser (may have been an addon for their 2dof pantograph lab kit, not sure), except that they are using a cable drive system rather then rack and pinion.

How long are you expected to spend on this project? is it a lab or a final uni project?

If your doing things like this over the internet you could look at kalman filtering or interpolation/prediction to compensate for unreliable communication latency.

You dont need any clever drive electronics, just a standard torque control servo amplifier. You ideally want the encoder and servo command signal to be sampled at 500hz-1khz on the pc, and to close the loop in software.

you dont need any physical springs, etc etc, you do that in software. As you know the position of both the devices, you can create a virtual spring force in software to apply to the user.

imagine using one device for now.

You start with the device at a rest state, displacement Xo. Now as you move the axis your position x changes. if you take the value d = X- Xo, you have a displacement of the axis. Now apply a force to the user, F, in such a way that it pushes against the users direction of travel.

F=Kd, where K is your chosen spring constant.

If you can do this you are half way there.


The next stage would be to change Xo so that it is set by the displacement of the second linear axis (actually its length - displacement for tug of war, so as one goes in the other comes out). This way as you push in the second axis, the first axis user feels force pushing them out.

The last stage would be to repeat the process on the other interface, so that both generate forces.