how to get force sensing on bottom of 3 DOF leg?

[NleahciM]

Hi - I'm working on a 6 leg, 3dof/leg machine. Pretty much everything is designed - except for one nasty little part. I'd like to have sensors on the legs that will tell me how much force is on the legs. This will help greatly in balance, and in knowing if the leg is slipping, etc. This is turning out to be a fairly nasty little bugger of a problem though. I should mention that everything is fairly small - the bottom leg segment is 5cm from center of rotation to tip, and is made out of milled 5mm thick aluminum. Also - I plan on having some sort of rubber tip on the legs. I will probabaly have to mold these rubber tips, so I'm very flexible when it comes to their shape.

One initial idea I had was to measure the current going to the motors (micro servos). But I think that that measurement technique would be clunky and really, really, really hard to do accurately.

So instead I've been looking at force sensors. I've been thinking I could do something along the lines of having the rubber tip have a shaft stick out of it, and then have a hole in the leg that goes straight down the middle of it, and then have the shaft slide inside that, with a force sensor at the end. Problem with this design is that it will only be accurate when the leg is at a known angle. (as when it's say at 45 degrees the force sensor will only have a part of the force on it). This could be solved by installing tilt sensors in each leg - but are there any tilt sensors that aren't fluid based (as those require fairly smooth movement I believe), are really small, and are relatively inexpensive? I haven't seen any.

So - what do you all think? Any ideas?

 

[williB]

One initial idea I had was to measure the current going to the motors (micro servos). But I think that that measurement technique would be clunky and really, really, really hard to do accurately.

what is the holding current of a micro servo..??
and what would you do with that info if you had it , unless your microbot was carrying something , fairly heavy , i dont see how that information wold help you if you had it..???

 

[DigiTan]

I was going to say "current measurement" until I got to the second paragraph. I'm guessing the solution would be clunky because of the A/D equipment needed on each leg? If it helps, I had a similar question that might be useful. I've heard a little about force sensors used in some 'robotic bug' projects, but I can't find any vendors.

Where did you get the micro servos, BTW? And how small are they? (I had questions about those too)

 

[NleahciM]

I was going to say "current measurement" until I got to the second paragraph. I'm guessing the solution would be clunky because of the A/D equipment needed on each leg? If it helps, I had a similar question that might be useful. I've heard a little about force sensors used in some 'robotic bug' projects, but I can't find any vendors.

Where did you get the micro servos, BTW? And how small are they? (I had questions about those too)

A/D is not the problem here. Accuracy is. Think of all the different situations:

A. servo is moving, no pressure on it
B. servo is moving, pressure on it
C. servo is not moving, no pressure on it
D. servo is not moving, pressure on it

And you think just with a magical A/D converter I could figure out which one of these is the case, and I could figure out how much pressure there is? No way. Not going to happen.

 

[NleahciM]

One initial idea I had was to measure the current going to the motors (micro servos). But I think that that measurement technique would be clunky and really, really, really hard to do accurately.

what is the holding current of a micro servo..??
and what would you do with that info if you had it , unless your microbot was carrying something , fairly heavy , i dont see how that information wold help you if you had it..???

You technically could monitor the current going into the servo and look for a spike, signifying that it was seeing resistance against it's movement. But as I said it's a bad idea. I have no idea of the actual numbers for the current, it obviously would differ alot between servos.

 

[tansis]

How about a piezo resistive sensor in a rubber foot or better still toes on the end of the leg? This tells you if the leg is in contact with the surface,plus it gives some indication of the orientation of the leg relative to the ground depending on how many toes you have and detects loss of traction.

 

[ice*]

You can try pressure sensors from Freescale Semi..
They got tiny ones.

They give a differential output,you just need an opamp..

 

[DigiTan]

A/D is not the problem here. Accuracy is. Think of all the different situations:

A. servo is moving, no pressure on it
B. servo is moving, pressure on it
C. servo is not moving, no pressure on it
D. servo is not moving, pressure on it

And you think just with a magical A/D converter I could figure out which one of these is the case, and I could figure out how much pressure there is? No way. Not going to happen.

Heh heh! No need to get nasty about it, chief! :lol: This is just a possibility. But you've got to admit it will fit your weight requirements and alert you if your motor stalls. The reason I bring it up is it's a common alternative to "dead recoking," and it becomes quite reliable once you've tuned the software to your servo.

It sounds like instead of a "pressure" or torque sensor, you're really wanting more an "offset sensor" to verify that your servos are really accelerating. Other than the Freescales ice* mentioned, my only suggestion is to use something like an MTS position sensor. This will probably require a 2D gimble depending on your range of motion, but they are very lightweight.

If you're still interested in leg angle measurement, I'd go with Analog's ADXL202-E. It works on 2 axes with 14 bits of resolution, plus it's 100% fluid free. But...they run about $18 each.

 

[NleahciM]

How about a piezo resistive sensor in a rubber foot or better still toes on the end of the leg? This tells you if the leg is in contact with the surface,plus it gives some indication of the orientation of the leg relative to the ground depending on how many toes you have and detects loss of traction.

My understanding of piezo stuff is that they only produce a voltage when a condition changes - thus you wouldn't know if a leg had just set down on something, or if it had just lifted up off of something. I would also think that finding the actual pressure would be a nasty task (I need more than just a contanct/no contact output - I need to know the balance of weight)

Interesting idea about the toes - though keep in mind the leg will probabaly be about 5-10mm wide... So I could only implement such an idea to a limited extent.

 

[NleahciM]

You can try pressure sensors from Freescale Semi..
They got tiny ones.

They give a differential output,you just need an opamp..

Hmm - some of those look very interesting - I'll have to spend a little more time there. Lately I've been looking at some made by CUI: https://www.cui.com/adtemplate.asp?invky=690700&catky=560054&subcatky1=895884&subcatky2= - I like them as they are very simple - just a basic variable resistor, they're very small, and they're fairly inexpensive and available (about $8 or so each and Digi-Key stocks them)

edit: looking through those - they seem to be for monitoring gas and fluid pressure - so I don't think they're what I'm looking for. Or am I missing something?

 

[NleahciM]

Heh heh! No need to get nasty about it, chief! :lol: This is just a possibility. But you've got to admit it will fit your weight requirements and alert you if your motor stalls. The reason I bring it up is it's a common alternative to "dead recoking," and it becomes quite reliable once you've tuned the software to your servo.

It sounds like instead of a "pressure" or torque sensor, you're really wanting more an "offset sensor" to verify that your servos are really accelerating. Other than the Freescales ice* mentioned, my only suggestion is to use something like an MTS position sensor. This will probably require a 2D gimble depending on your range of motion, but they are very lightweight.

If you're still interested in leg angle measurement, I'd go with Analog's ADXL202-E. It works on 2 axes with 14 bits of resolution, plus it's 100% fluid free. But...they run about $18 each.

For measuring angle - I've been considering trying to use the signal from the potentiometers in the servos. But those piher pots look pretty nice if I end up not doing that for whatever reason.

I think I've been unclear in what I want the force sensors for - I want to know how weight is balanced. I also want to know if the ground surface changes, so that the machine can move its legs to adjust. Sorry for not being clear in the matter.