3D positioning of a hand held device.

[kaarvannan]

Hi all,
I have a rather unique requirement with regards to position sensing. I need to sense the position of a hand held device (kinda like the wii controller) in a 1^3 meter volume with very high accuracy (in micro meters). The device will be picked up and moved by hand, so encoders will not work. I was thinking if there is way to transmit somthing (IR, elctromagnetic waves or Ultrasonic waves over this 1^3 volume then, i can place a receiver or reflector of some sorts in the hand held device and make it work.
On the plus side, we can run wires to the hand held device, thus wireless communication is not needed and the whole thing will be controlled by labview or matlab running on a PC so there is a lot of computing power.
I am open to any idea you might have to accomplish this?

 

[duffy]

μm? 1/1,000,000th of a meter? For that kind of resolution you are going to need something like laser interferometry, things like field coils and ultrasonic transducers just aren't going to cut it. If you are trying to make a 3D digitizer with a hand-held stylus, you are going to also need to resolve it to a fraction of an arc-second rotationally.

 

[kaarvannan]

thats right! the resolution I need is 0.001 inch which is 25.4 micro meters. Can a laser cover the 1 cubic meter volume I had specified?
I was thinking along the lines of a magnetometer in a magnetic field. However, I am yet to find anything with the range and precision I am looking for.
And Yes, what I am trying to do is similar to a 3D digitizer with a hand-hand stylus. I dont understand what you mean by

resolve it to a fraction of an arc-second rotationally

? can you post some links about this topic, for me to read-up.
Thanks.

 

[duffy]

I don't know of any links, but this topic comes up from time to time. "Resolve it to a fraction of an arc-second rotationally" is in reference to the fact that you have to calculate an offset from the sensor to the point of contact, which can be at an angle - say it was like a pencil, and your sensor is where the eraser is, but what you are calculating is where the pencil-point is; you need to know the angle the pencil is sitting at. (also the distance between the two, but that's a fixed number)

 

[duffy]

Here - take a look at this:
https://polhemus.com/?page=Motion_PATRIOT Digitizer

Sadly, their best accuracy is only about .028"
https://www.electro-tech-online.com...ompare20with20ResolutionvsRange20graphs-8.pdf

 

[cr0sh]

Here - take a look at this:
https://polhemus.com/?page=Motion_PATRIOT Digitizer

Sadly, their best accuracy is only about .028"
https://www.electro-tech-online.com...ompare20with20ResolutionvsRange20graphs-1.pdf

I would almost say that if you can't get it from Polhemus or Ascension, you probably can't get it at all; they have been the two major players in 3D tracking and sensing for over 20 years. Furthermore, if you could get the accuracy wanted, you'll probably spend an arm and a leg (and a piece of your soul) on it. Just to give an idea: A few years back I made a "steal" on Ebay; I picked up an older RS-232/422 Ascension FOB kit - dual sensors, with the smaller "desktop" cube (thing weighs a ton!). I paid something like $300.00 USD for this bit of kit.

I contacted Ascension about it, mainly asking about the cost to get either the large area tracking cube, or the "pad" tracker transmitter they had; in the course of the conversation, I asked about how much my system would cost if purchased "today". I got back a figure of around $6K USD! That would be for the -exact- same system (at the time, they sold USB versions, and a lot of other "new products"), which was already around 12-15 years old. It certainly doesn't have anywhere near the accuracy the OP is wanting (and I am not sure it even has the .028" accuracy).

I'm guessing that likely such a 3D tracking device doesn't exist - or if it does, it won't come cheap - not cheap at all...

 

[kaarvannan]

After doing some research I have refined my accuracy expectations to realistic levels. cr0sh is right, it is really expensive to do what i am trying to do. Ascension's laser based optical systems which can do this cost a little less that $20,000!
So I have decided to do this - use an IMU to sense the tilt and measure the distance of the hand held device from a base station. So, with three angles and a distance I can find the position of any point with respect to the base station(using some constraints in my application). However, this leads me to my next issue.
I need a way to measure the distance of a point from the base station. I can not use anything optical, since there are line of sight issues. I can not use a magnetic sensing since there will be large metal objects nearby. So I am pretty much left with ultrasonics, but there are issues with sensing angle (angle between the transmitter and receiver). Can you think of anything that will give me the distance between a transmitter, receiver even when the receiver is turned, say 90 degrees.

 

[duffy]

A gyro's not a bad idea for measuring angles, though there are a few caveats: if you tap it and it can throw it off because it exceeds the rate, and the angular resolution and linearity suggest putting it as close to the buisness end of your device as you can.

an IMU to sense the tilt and measure the distance of the hand held device from a base station.

Using an accelerometer to measure distance is very inaccurate. What you are doing is integrating acceleration over time to find velocity, then integrating velocity over time to find distance. This magnifies the error. We experimented with one of these in this way as part of a map project at my old job, the thing just never came back to the same spot after walking around with it.

 

[kaarvannan]

I am not going to use accelerometer to find the position. I am planning to use a IMU (which fuses an accelerometer and a gyro in the same chip to overcome drift errors) to find the tilt. I will also measure the distance of the hand held device from the base station using ultrasonic transducer-receiver combo. this will be sufficient for my project.

 

[duffy]

You said you can't use magnetic sensing because of the metal objects you are working with, but ultrasonics will reflect off surfaces, taking longer paths that could give false readings as well. Could you make a combination of the two? One backs up the other, if it doesn't actually enhance resolution it might at least detect that there is disagreement and make the reads more reliable.

 

[alec_t]

With ultrasonics you will also have to allow for variations in the speed of sound with air density and temperature.