Hi all,
I am looking for a way to sense the position of a person's feet in an area of a square metre or so. It need to have a high enough resolution to determine in which way each foot is oriented, but it does not need to be phenomenally accurate. Some degree of pressure sensitivity would also be desirable. I am basically looking for a large, low-resolution multi-touch sensor that is robust enough to be trodden on without breaking. It should also be as thin as possible!
I initially thought of doing this using the capacitance between wires in a grid with some kind of squishy resistive medium in between. Reading the capacitance at each crossover of wires would give an indication whether or not a foot is standing there. Has anybody got any experience of building a similar system, even if it is on a smaller scale? Also, there are systems which work using the capacitance of the human body (like the iPhone screen) - this has the advantage of being thinner and not requiring any kind of squishing. Would this be difficult to scale up?
I also thought of those terrible dancing games you can buy for consoles. I seem to recall that they work using flexible mats with a pressure pad under each 'button'. Has anybody ever dissected one of these? What kind of pressure sensor is flexible and thin enough to be rolled up?
Any help greatly appreciated, I will be sure to share any successes with the forum!
Cheers,
Dan
I think this is a much harder problem than you think. The game type foot pad is just a simple strip of foil held apart by a bit of foam, it hasn't got the resolution you need.
To be able to detect the way a foot is pointing requires sensors every half inch or so over the area of the mat. That's a lot of sensors and wiring.
You could make a scaled up version of a touch screen with resistive plastic, the sort of thing electronic boards are packed in but that would not be multi touch, which is what you want.
Maybe the best bet would be to use some sort of web cam and image processing on your main computer rather than an Arduino.
This might be to difficult, and too small, if you could get the data via bluetooth from the wii balance board. I have thought of this but never new anyway to do it. You could probably connect it some how and you should look into an idea like this....
Problem is that the wii does not sense foot position like he wants. It only senses weight distribuition. Therefore there will be no way of determining what way round the foot is pointing.
Mike - In reference to requiring sensors every half inch or so - I'm not sure this level of detail would be altogether necessary; I suppose the average foot is 8-9 inches long or so, and I only really need to know its orientation in a very basic way (ie whether it is 0, 45 or 90 degrees to the Y axis of the sensor platform). I think that something like this might still work - however, it wouldn't give me any pressure sensitivity.
Concerning the use of a more touchscreen-esque technology, I had thought of dealing with the 'multitouch' problem by splitting the sensor area into strips. Imagine the sensor platform to be two feet long and a foot wide. There could be six one-inch-wide strips of sensor material running the whole length of the platform, each an inch apart (or something similar). This would give lots of Y resolution, which is more important for my particular problem, and enough X resolution to get along with.
By resistive plastic, do you mean those silver anti-static bags? Using this as the dielectric of the capacitor? Or do I have the wrong end of the stick?
Re: the Wii thing - the problem with using anything other than a touchscreen/pad type thing is that I need this to be very flat, and flexible too.
By resistive plastic, do you mean those silver anti-static bags?
No not the silver ones the thick soft black plastic ones you get about 20 to 50 K over a couple of inches. The silver plastic ones don't conduct at all.
I was thinking you needed foot positions of 180 degrees in which case you would need a higher resolution than you appear to.
One solution might be a lattice, horizontal strips of adhesive backed copper tape on a top sheet with vertical strips on the bottom sheet, sandwiched together with conducting foam. You can get your resolution with that if you scan them like a switch matrix, but you would need several of these if you want the multi touch.
How can you detect multiple positions if the X,Y if the layers of "detection material" (whatever it may be) overlap? Unless you have a dedicated right hand side and left hand side.
If separate sensors were used, it would be like multi-touch without being multi-touch. Imagine a stripy lawn, each stripe being a different touchpad. You can detect two separate feet as long as they're not both on the same strip - which for this application would be unlikely.
Grumpy - I'm still not sure of the material you mean, could you provide a link?
Ideally I am thinking along the lines of a grid of capacitive touch sensors - each one just sensing on/off, but with enough of them to give some decent resolution. These would allow my system to be extremely thin, since a sensor like this will work with thin copper traces. I found this page, which looks useful - http://www.planetanalog.com/features/showArticle.jhtml?articleID=181401898. They talk about sensing through a 1cm glass overlay, I wonder what kinds of plastic would be suitable as a dielectric.
I think there are a few unstated issues you're not considering:
First, there's the issue of how to tell which end is the toe and which is the heel. Many types of shoes are nearly symmetrical is this regard, so it might be very difficult to tell which end is the toe even if you could capture a complete scan of the sufaces in contact with your pad.
Even if you had a very fine grid of pressure sensors and tried to find the toe end by seeing which end pressed down with ledd weight, this would still change as someone shift weight back and forth while walking.
One thing that might work would be to have a two level array of optical beam sensors (infraread LEDs on one side and phototransistors on the other.) One level would be at heel height and the other at ankle height. Using this, you might be able to ciscern foot orientation comparing the points of intersection detected by these arrays.
It might be hard to deal with the issue of one foot shading the other from the scanners, but good detection algorithm might be able to deal with this. Just a thought.