Post by thread:[EE] Methods for tracking finger position

I have been tasked with doing a conceptual design for a device that tracks finger movement using a sensor device mounted to the finger. Does anyone have experience with this or something similar? Or thoughts or ideas? For now, I only have to worry about one finger - but need to track position in 3 dimensions with a resolution of about 5mm over an area of 1 cubic meter. Also, the device cannot use a reflective method (such as ir or ultrasonic). So far, I have investigated using accelerometer chips, but these are still fairly expensive and I would like a lower-cost alternative. Thanks, Randy

Randy Boulter wrote: > > I have been tasked with doing a conceptual design for a device that tracks finger movement using a sensor device mounted to the finger. Does anyone have experience with this or something similar? Or thoughts or ideas? For now, I only have to worry about one finger - but need to track position in 3 dimensions with a resolution of about 5mm over an area of 1 cubic meter. Also, the device cannot use a reflective method (such as ir or ultrasonic). So far, I have investigated using accelerometer chips, but these are still fairly expensive and I would like a lower-cost alternative. > > Thanks, > > Randy How about strings from 3 different points that are all tied to the finger via pulleys. You can use the position of the pulleys to calculate the finger position. - Marcel

>For now, I only have to worry about one finger - but >need to track position in 3 dimensions with a resolution >of about 5mm over an area of 1 cubic meter. Also, >the device cannot use a reflective method (such as ir >or ultrasonic). So far, I have investigated using >accelerometer chips, but these are still fairly expensive >and I would like a lower-cost alternative. I suspect that accelerometers will have long term drift. What sort of long term accuracy do you need? One thought I had was to use one of those sensory gloves that are used for virtual reality type things. You would need to have the person sitting to get a reference point, or some other method of reference.

----- Original Message ----- From: "Marcel Birthelmer" <spam_OUTmarcelb.listsTakeThisOuTgmail.com> To: "Microcontroller discussion list - Public." <.....piclistKILLspam@spam@mit.edu> Sent: Wednesday, May 31, 2006 12:16 PM Subject: Re: [EE] Methods for tracking finger position > How about strings from 3 different points that are all tied to the > finger via pulleys. You can use the position of the pulleys to calculate > the finger position. > - Marcel > -- This would be great for one finger, which is what I said I needed. But the application will eventually need to track 10 fingers, so I need to keep this in mind. Thanks, Randy

> I have been tasked with doing a conceptual design for a device > that tracks finger movement using a sensor device mounted to > the finger. Does anyone have experience with this or something > similar? Or thoughts or ideas? For now, I only have to worry > about one finger - but need to track position in 3 dimensions > with a resolution of about 5mm over an area of 1 cubic meter. > Also, the device cannot use a reflective method (such as ir or > ultrasonic). So far, I have investigated using accelerometer > chips, but these are still fairly expensive and I would like a > lower-cost alternative. I'm not certain this is a feasible project within the parameters given. Particularly, the fact that you're feeling accelerometers are too costlly gives me pause. If you can't handle US$6-12 per finger, which is the cost of many MEMS accelerometers these days, you are not likely to find a non-reflective method for tracking ten fingers for less than that. That said, I believe one of the ways VR gloves do their thing is by using strain-sensitive or flex-sensitive resistors to measure the degree to which the fingers are bent. Not sure that's what you need. Are you trying to track the tip of the finger, then, or the whole finger, so that a model of the finger can be reconstructed later? Mike H.

Randy Boulter <wrb <at> isp.com> writes: > > > I have been tasked with doing a conceptual design for a device that tracks finger movement using a sensor > device mounted to the finger. Does anyone have experience with this or something similar? Or thoughts or > ideas? How about using a (ultra)-sound source at the finger and 3 microphones to triangulate the position. Something like this was used for DNA sequence readout from autoradiographs using a pen as pointing device. The resolution was in sub-mm range (IIRC). Another option is to use 3 sound transducers and microphone on the finger to find position, similar to GPS, only with ultrasound. Sergey

Alan B. Pearce wrote: >>For now, I only have to worry about one finger - but >>need to track position in 3 dimensions with a resolution >>of about 5mm over an area of 1 cubic meter. Also, >>the device cannot use a reflective method (such as ir >>or ultrasonic). Why not? Ultrasonic emitters on the tips of the fingers would be tiny and using time-of-flight measurements, and 4 receivers, easy to decode. Use multiple frequencies if you need simultaneous multiple tracking. What sample rate do you need? >> So far, I have investigated using >>accelerometer chips, but these are still fairly expensive >>and I would like a lower-cost alternative. HUH? $25 for 3 axis 1G (ADXL 303JE) is pretty darned cheap. But they probably won't give you the kind of accuracy you want. And you've said nothing about tracking orientation. > I suspect that accelerometers will have long term drift. What sort of long > term accuracy do you need? > > One thought I had was to use one of those sensory gloves that are used for > virtual reality type things. You would need to have the person sitting to > get a reference point, or some other method of reference. These gloves use magnetics, (Flock of birds, 6 DOF) for gross position and fibre optics to detect bending of the fingers. Stay away from the Polhemous stuff. It's AC and suffers from big issues with reflections (eddie currents). Robert

Sergey Dryga wrote: {Quote hidden}> Randy Boulter <wrb <at> isp.com> writes: >> >>I have been tasked with doing a conceptual design for a device that tracks > > finger movement using a sensor > >>device mounted to the finger. Does anyone have experience with this or > > something similar? Or thoughts or > >>ideas? > > > How about using a (ultra)-sound source at the finger and 3 microphones to > triangulate the position. Something like this was used for DNA sequence > readout from autoradiographs using a pen as pointing device. The resolution was > in sub-mm range (IIRC). > > Another option is to use 3 sound transducers and microphone on the finger to > find position, similar to GPS, only with ultrasound. This method works quite well. We had an old 3-d digitizer that used a pen that snapped (electric arc) and 3 mics with about a 0.5 meter spacing, and it covered the 1/2 meter cube quite well. With 4 external sources, (to get better closure on the RMS position error), and phase measurements, one could get very good accuracy, if a continuos high sample rate was required. Sounds to me like you're looking to do performance art, so there is no budget available to use the optimal technology. Robert

Forgive me for not providing a more complete description. The intended applicaton is a type of "virtual keyboard". For now, I need to simulate, using a finger-mounted sensor, the action of reaching out and touching a key with one finger. And then be able to track the finger as it is lifted and moved to another key. This will eventually be expanded to 10 fingers. I located some information on the sensors you mentioned - these could be used to determine if the finger is bent (i.e pressing a key), but that still leaves the problem of locating the finger in space. BTW, $6-12 per finger is not a show-stopper - I would just like to do it for less if possible. Thanks, Randy {Quote hidden}> > I have been tasked with doing a conceptual design for a device > > that tracks finger movement using a sensor device mounted to > > the finger. Does anyone have experience with this or something > > similar? Or thoughts or ideas? For now, I only have to worry > > about one finger - but need to track position in 3 dimensions > > with a resolution of about 5mm over an area of 1 cubic meter. > > Also, the device cannot use a reflective method (such as ir or > > ultrasonic). So far, I have investigated using accelerometer > > chips, but these are still fairly expensive and I would like a > > lower-cost alternative. > > I'm not certain this is a feasible project within the parameters given. > Particularly, the fact that you're feeling accelerometers are too > costlly gives me pause. If you can't handle US$6-12 per finger, > which is the cost of many MEMS accelerometers these days, you > are not likely to find a non-reflective method for tracking ten fingers > for less than that. > > That said, I believe one of the ways VR gloves do their thing is by > using strain-sensitive or flex-sensitive resistors to measure the > degree to which the fingers are bent. Not sure that's what you need. > > Are you trying to track the tip of the finger, then, or the whole finger, > so that a model of the finger can be reconstructed later? > > Mike H. > -

Look at the Nintendo Wii's controller. It uses an infrared LED in the controller and for a base station has two cheap IR cameras a certain distance apart. Position is read off the cameras and distance is measured using triangulation. This method seems to give pretty accurate absolute positioning for a very cheap price point. Cheers, Zik On 01/06/06, Randy Boulter <wrbKILLspamisp.com> wrote: > > > I have been tasked with doing a conceptual design for a device that tracks finger movement using a sensor device mounted to the finger. Does anyone have experience with this or something similar? Or thoughts or ideas? For now, I only have to worry about one finger - but need to track position in 3 dimensions with a resolution of about 5mm over an area of 1 cubic meter. Also, the device cannot use a reflective method (such as ir or ultrasonic). So far, I have investigated using accelerometer chips, but these are still fairly expensive and I would like a lower-cost alternative. > > Thanks, > > Randy > -

> Sergey Dryga wrote: > > > Randy Boulter <wrb <at> isp.com> writes: > >> > >>I have been tasked with doing a conceptual design for a device that tracks {Quote hidden}> > > > finger movement using a sensor > > > >>device mounted to the finger. Does anyone have experience with this or > > > > something similar? Or thoughts or > > > >>ideas? > > > > > > How about using a (ultra)-sound source at the finger and 3 microphones to > > triangulate the position. Something like this was used for DNA sequence > > readout from autoradiographs using a pen as pointing device. The resolution was > > in sub-mm range (IIRC). > > > > Another option is to use 3 sound transducers and microphone on the finger to > > find position, similar to GPS, only with ultrasound. Robert Rolf wrote : {Quote hidden}> This method works quite well. We had an old 3-d digitizer that used > a pen that snapped (electric arc) and 3 mics with about a 0.5 meter > spacing, and it covered the 1/2 meter cube quite well. > > With 4 external sources, (to get better closure on the RMS position > error), and phase measurements, one could get very good accuracy, > if a continuos high sample rate was required. > > Sounds to me like you're looking to do performance art, so there is > no budget available to use the optimal technology. > > Robert This sounds like an excellent method, but how far apart would the microphones have to be if they were placed, say 20 feet away from the sound source (finger), and still maintain accuracy within a few millimeters? Randy

When you come up with a method for $12 per finger, let me know I would love to help you test it. POD 0000000000000000000000000000000000000000000000 On 2006-May 31, at 17:08hrs PM, Randy Boulter wrote: Forgive me for not providing a more complete description. The intended applicaton is a type of "virtual keyboard". For now, I need to simulate, using a finger-mounted sensor, the action of reaching out and touching a key with one finger. And then be able to track the finger as it is lifted and moved to another key. This will eventually be expanded to 10 fingers. I located some information on the sensors you mentioned - these could be used to determine if the finger is bent (i.e pressing a key), but that still leaves the problem of locating the finger in space. BTW, $6-12 per finger is not a show-stopper - I would just like to do it for less if possible. Thanks, Randy {Quote hidden}>> I have been tasked with doing a conceptual design for a device >> that tracks finger movement using a sensor device mounted to >> the finger. Does anyone have experience with this or something >> similar? Or thoughts or ideas? For now, I only have to worry >> about one finger - but need to track position in 3 dimensions >> with a resolution of about 5mm over an area of 1 cubic meter. >> Also, the device cannot use a reflective method (such as ir or >> ultrasonic). So far, I have investigated using accelerometer >> chips, but these are still fairly expensive and I would like a >> lower-cost alternative. >> > > I'm not certain this is a feasible project within the parameters > given. > Particularly, the fact that you're feeling accelerometers are too > costlly gives me pause. If you can't handle US$6-12 per finger, > which is the cost of many MEMS accelerometers these days, you > are not likely to find a non-reflective method for tracking ten > fingers > for less than that. > > That said, I believe one of the ways VR gloves do their thing is by > using strain-sensitive or flex-sensitive resistors to measure the > degree to which the fingers are bent. Not sure that's what you need. > > Are you trying to track the tip of the finger, then, or the whole > finger, > so that a model of the finger can be reconstructed later? > > Mike H. > -

Randy Boulter wrote: {Quote hidden}> > Robert Rolf wrote : >> We had an old 3-d digitizer that used >>a pen that snapped (electric arc) and 3 mics with about a 0.5 meter >>spacing, and it covered the 1/2 meter cube quite well. >> >>With 4 external sources, (to get better closure on the RMS position >>error), and phase measurements, one could get very good accuracy, >>if a continuos high sample rate was required. >> >>Sounds to me like you're looking to do performance art, so there is >>no budget available to use the optimal technology. >> >>Robert > > > This sounds like an excellent method, but how far apart would the > microphones have to be if they were placed, say 20 feet away from the sound > source (finger), and still maintain accuracy within a few millimeters? First you say 1 meter (3 feet) and now you're asking for 20 feet. BIG change in specifications here. Pointless for me to waste my time suggesting solutions for a moving target. The accuracy would be determined by noise levels, and how accurately you measure the transit time. Robert