37
\$\begingroup\$

I am dealing with a situation where a patient can only very lightly press his fingers / hand to signal what he / she wants. Speaking or eye tracking is not possible.

The person is also not able to lift his hand. The only movement possible is a slight press with the hand / fingers.

I was therefore thinking about using a capacitative device that would be calibrated to detect how much pressure is a willing / intended press and which amount pressure is simply a resting position of the hand on the device.

I guess a capacitative device is what is required here. I saw a solution of a custom device here on YouTube, but that looked so custom-made and so rigid that I was wondering if what I need actually already exists.

I would therefore like to ask how I could measure the pressure of a human finger / hand on a device in a reliable way.

Thank you.

Edit: Nobody so far suggested to use a precision balance with USB output option. May I ask if that would not work?

enter image description here

\$\endgroup\$
12
  • \$\begingroup\$ So the person can lift finger? But not their hand? Can they drag their hand side to side? \$\endgroup\$ – DKNguyen May 13 at 0:30
  • 2
    \$\begingroup\$ @DKNguyen The only reliable gesture that the patient can do is a press with the hand. So if you hold his hand, and he presses, you will notice the pressure clearly. But wide movements are not possible. So for example pressing a button is not possible. \$\endgroup\$ – tmighty May 13 at 0:33
  • \$\begingroup\$ Can they press with individual fingers? \$\endgroup\$ – DKNguyen May 13 at 0:34
  • 2
    \$\begingroup\$ Strain gauge? researchgate.net/publication/… \$\endgroup\$ – Mattman944 May 13 at 0:44
  • 1
    \$\begingroup\$ I am looking at laser fiber optic whiskers described in Gordon McComb's book which might be a bit more reliable than capacitive pressure sensing if you need to detect degree and less picky about materials and form factor. For example, you could embed the fibre in a soft deformable plate. Oh, by the way I just noticed your post technically counts as a product recommendation which is off-topic here. \$\endgroup\$ – DKNguyen May 13 at 0:47

10 Answers 10

38
\$\begingroup\$

There is a component called a "force-sensitive resistor" that I'd suggest you try. You can get them in a small, round configuration like this one from Adafruit force-sensitive resistor. enter image description here

They are easy to read as they just appear as a variable resistance; as the finger pressure is increased, the resistance decreases. Make a voltage divider with the FSR as one leg, then most any microcontroller with an analog input can measure the pressure.

You can also make these yourself using a sandwich of piezoresistive sheet between two layers of conductive fabric, but the ready-made FSR is easier.

A capacitive sensor may also work, but detecting variation in pressure in a consistent way is trickier. The capacitance will increase with finger pressure, but the capacitance will also be affected by the skin's hydration, surface moisture, etc.

\$\endgroup\$
3
  • 1
    \$\begingroup\$ This was going to be my suggestion, good to see someone got there first! I've used these a few times myself. Very simple, easy to make solution \$\endgroup\$ – MCG May 13 at 8:50
  • \$\begingroup\$ Yep, thats what I was gonna recommend. \$\endgroup\$ – Randomaker May 13 at 12:27
  • \$\begingroup\$ I don't think absolute pressure is needed, so a capacitive sensor could be continuously recalibrated on a time scale of around 5 to 10 seconds, so that a reduction or increase in pressure lasting a few seconds would be reliably detected. \$\endgroup\$ – Nobody May 13 at 19:46
31
\$\begingroup\$

I've tried out a bunch of different pressure sensing technologies for robotic manipulation (specifically robot prosthetics). There are many that are easy to manufacture with limited resources (3d printing + cheap PCBs + off the shelf, room temp cure silicone rubber). One of the best sensors I ever made was based on this paper. It's quite literally a blob of silicone rubber that suspends a magnet over a 3-axis hall effect sensor (the one in the paper is the MLX90393, but there are many such devices on the market that would work equally well). It's extremely simple and cheap to make, highly reliable, sensitive, and can measure shear forces as well.

I ultimately wasn't able to use this for my robot manipulator, because magnetized objects (i.e. screwdrivers, any steel object with light magnetization) caused false positive pressure readings. However, since you'll be interfacing only with a patient's finger, stray field lines from other objects won't really be an issue. The extra dimensions of sensitivity also may allow you to maximize whatever range of motion is left in your patient. You also should be able to 'tune' the sensitivity by changing the stiffness of the silicone blob. You'll want to purchase platinum cure silicones that are 'skin safe' for your application as well.

Link to video showing off 'feather touch' and light finger touches.

\$\endgroup\$
7
  • 4
    \$\begingroup\$ That's clever! But word of caution: room-temp cure, off-the shelf silicone rubber releases acetic acid (you can smell it strongly as it cures). I'd worry about putting it in prolonged contact with a frail patient's fingertip. But there are more expensive, platinum-cure silicone rubber compounds that would be skin-safe. \$\endgroup\$ – Mark Leavitt May 13 at 15:36
  • 5
    \$\begingroup\$ We used 'dragon skin' from smooth-on, which is platinum-cure and has a 'skin-safe' certification. You can get a 2lb 'trial unit' for ~$30, which is enough to easily make 5-10 of these. Definitely a good point though, buying specifically skin-safe silicones is important for this. \$\endgroup\$ – Ocanath May 13 at 15:41
  • \$\begingroup\$ @Ocanath Thank you very much. That sounds really interesting. Are you the author of the YouTube video? I am asking because I would be really interested to test / use this clever solution. But I have too little electronic experience to build it myself, left alone the USB connection, etc. I would so much like to buy a ready-made device. \$\endgroup\$ – tmighty May 13 at 16:21
  • \$\begingroup\$ I'm not the author but I've replicated the work. I've made a little dome sensor like the one in the video, and integrated an array of these sensors in a robot prosthetic finger prototype (which we ultimately didn't use, due to the stray field issue I mentioned). There are sparkfun breakouts for this chip you can buy off the shelf and hook up to an arduino (Qwicc), which means the libraries for getting data and streaming it to a PC are easy to use. (I'd be happy to provide more specific help, feel free to reach out to me via. contact info in my profile) \$\endgroup\$ – Ocanath May 13 at 17:14
  • \$\begingroup\$ @Ocanath Would you please edit your notes about the skin-safe silicones into your answer? Comments are transient. \$\endgroup\$ – chrylis -cautiouslyoptimistic- May 14 at 6:34
11
\$\begingroup\$

You could take a look at fiber optic laser whiskers. What you do is you get a laser diode and send it down a cheap, plastic, crappy, step-graded fiber optic loop (you don't want the expensive continuous graded or glass kind) with a photodiode on the other end.

Deforming the loop, even slightly changes the the interference pattern read by the photodiode or phototransistor. The output should correspond with how badly the loop has been deformed.

It might be possible to use one of the laser diodes with an integrated photodiode to send the signal down a straight fiber with a reflector of some sort on the other end. That way you wouldn't need a loop. I don't really understand the way the decoherence and interference pattern works in fiber so I don't know if light taking the same path down the wire both ways would still produce the changes you need (they might cancel out and eliminate the changes you need or they might double up and make things more sensitive, I don't know).

You can then put it in any cushion or substrate you want, and a hard plate on top of that if you wish. Or make a glove or put it in a silicone or plastic tube that the guy can squeeze, or whatever other form factor you want.

I've never used this for anything myself but it is written about in Gordon McComb's book Robot Builder's Bonanza but you could probably also find other search results about it. In the book he says they are very, very sensitive. Feather touch is how it is described.

There are also Fiber Bragg Grating sensor which are smaller and simliar but not identical principles. You'd need to buy these as is since they are made with a specially constructed fiber. These are what is used on commercially available optical strain gauges (as opposed to more conventional metal alloy resistive strain gauges). They sound really cool.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ You might get enough reflection from a cleaved end (just knife-cut if a plastic fibre, which will also deform more easily than glass). If the coupling into the cladding is what changes with pressure, you won't get that back on the return trip, so reflection could work \$\endgroup\$ – Chris H May 13 at 8:53
7
\$\begingroup\$

I would try making device where is LED shining at some angle to photoresistor. When the LED (or photoresistor) changes the angle, the amount of light on the photoresistor decreases or increases.

LED-moving variant:

example construction

Photoresistor-obscuring variant:

example construction

Similar device can be also made with fixed LED and photoresistor, when some opaque part of the device partially obscures the photoresistor.

Such devices should not have any problems with signals from the environment (such as magnetism, electrical charges etc.). The device should be lightproof and the LED should have stable power supply.

\$\endgroup\$
2
  • \$\begingroup\$ While I suspect this isn't the best solution for the OP, it still gets a +1 from me as it would be good in other cases (e.g. a student project) that doesn't have to last so long. Careful mechanical design could be used to 3d-print a housing for the LED and photodiode, that also provided the flex, all in one part - but IME 3d-printing flexing parts requires a bit of trial and error \$\endgroup\$ – Chris H May 14 at 7:44
  • \$\begingroup\$ This should be extremely inefficient in every category. Besides, damper will be required here. And if you use moving parts, inductor with moving core or capacitor with moving plate is better solution. \$\endgroup\$ – Sadat Rafi May 15 at 10:59
5
\$\begingroup\$

An alternative pressure sensor is the Quantum Tunnelling Composite materials from Peratech which changes its resistance from megohms through kilohms to ohms according to the applied pressure.

A good article about the technology here

Small "pills" of this material used to be available here or here (but apparently no more) Maplin in the UK also used to sell them, but no more.

It might be easier to scan a 2D array of these looking for low resistances than detecting voltage from a large array of piezo sources.

Wikipedia mentions licensing deals with another manufacturer but their website appears to be dead atm..

\$\endgroup\$
5
\$\begingroup\$

An alternative is to measure the patient's muscle instead of their hand movement. You can get muscle sensors that connect to electrodes that you'd stick to the patient's bicep, brachioradialis, or whatever muscle they're using to apply pressure with. When they press with their hand, you detect the muscle flexing instead of the pressure. Depending on how sensitive your sensors are, you might even be able to detect movements in other muscles that are controllable by the patient, but are too small to have any visible effect on hand movement/pressure.

\$\endgroup\$
5
  • \$\begingroup\$ Amazing, thank you, I will try that. \$\endgroup\$ – tmighty May 15 at 0:45
  • \$\begingroup\$ This wouldn't work as the muscle intensity so minimal. The myoware is intended for really strong muscles tensions as far as I can judge from its promo video which shows a bodybuilder flexing his biceps. \$\endgroup\$ – tmighty May 15 at 12:29
  • 1
    \$\begingroup\$ There are prosthetic limbs that are controlled by muscles on the stub, so it should be possible for this to work \$\endgroup\$ – CSM May 18 at 8:18
  • \$\begingroup\$ no, this is not a good option. EMG is not an option for a paralyzed patient since paralysis means there's no EMG to pick up. \$\endgroup\$ – Ocanath May 20 at 1:07
  • \$\begingroup\$ @tmighty That particular sensor kit is more hobbyist-grade, so I'm not too surprised if it isn't very sensitive (I couldn't find any explicit specs). Medical-grade sensors, however, are capable of reading absurdly low levels of activity. Like most things, it's all a big trade-off between how much sensitivity you need and how much you want to pay. \$\endgroup\$ – bta May 20 at 19:25
4
\$\begingroup\$

This would probably require testing and calibration, but pressing the finger does deform it slightly, increasing contact area, which in turn increases the measured capacitance.

I am not sure whether this difference is enough to be detectable and would most likely require per-patient calibration, but it is something you might want to check.

\$\endgroup\$
1
4
\$\begingroup\$

The classic solution here is a strain gauge bonded to the side of a metal or hard plastic grip. They're accurate, cheap, and easy to interface to. They're used in all manner of sensors for exactly this kind of application.

Capacitive sensors are very, very accurate, but by orders of magnitude more than you need for this. (I actually work for a company which makes capacitive sensors - some kit made by my company back in the day went into the Canadarm2 on the ISS for strain measurement). They aren't cheap, and nor is the electronics to drive it.

Touchscreens do use capacitive sensors too, of course, but they're designed as a binary sensor. Although you can do some stuff with them for analogue measurement, they're not a great place to start.

\$\endgroup\$
3
  • \$\begingroup\$ Can you search the pieces for me that I would need to build and connect this strain gauge to a Windows computer where I can analyze it? I know it's a curious question, but I am really newbie at this still. \$\endgroup\$ – tmighty May 15 at 11:32
  • \$\begingroup\$ @tmighty If you google "strain gauge kit" you'll get a fair amount of options. For connecting to a PC, you could start by putting it into the headphone socket. If you google "PC headphone oscilloscope" you'll get details of that too. \$\endgroup\$ – Graham May 15 at 17:38
  • \$\begingroup\$ @tmighty Sorry, I wasn't quite right there. The headphones of course would be for output. For input, you want your PC's audio line in. Google "pc sound card oscilloscope". Laptops generally don't have a line in, but you can get a cheap USB adapter. \$\endgroup\$ – Graham May 15 at 21:07
3
\$\begingroup\$

Look at a strain gauge, they make some very small ones that should work. You could also adjust the sensitivity over time or a learning method and as the patient improves so does the force required.

\$\endgroup\$
2
\$\begingroup\$

There are lots of ways to measure pressure, but it seems like it might work better to just use a leaf-type microswitch, or some similar switch that clicks when it engages and disengages. This won't distinguish degrees of pressure, but it seems likely that a patient in this state wouldn't be able to reliably control multiple pressure levels.

If the patient has any sense of touch left, haptic feedback from a click-switch can make a huge difference in usability. The amount of pressure required can be tuned by adjusting lever length or position.

\$\endgroup\$
4
  • \$\begingroup\$ This doesn't work because the patient doesn't have enough power to lift his hand after having pressed the trigger. So lifting is even more difficult that building up some pressure. \$\endgroup\$ – tmighty May 15 at 0:43
  • \$\begingroup\$ the switch could be mounted in a rod that is placed in their hand \$\endgroup\$ – Jasen May 15 at 4:28
  • \$\begingroup\$ @Jasen Hmmm, it's really super limited. You put his hand into yours, and you can feel a feather light press. There is literally no movement possible, only a press. \$\endgroup\$ – tmighty May 15 at 9:17
  • \$\begingroup\$ @tmighty These switches are spring-loaded, so if a feather-light press activates it, removal of that feather-weight will deactivate it. \$\endgroup\$ – jeffB May 15 at 17:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.