Measure a <X,Y,Z> point within 0.1mm over 10m x 10m x10m

[terryphi]

Hello,

I'm looking for some ideas as to how I can get the position of a beacon/actuator to within 0.1mm in a cube 10m x 10m x 10m.

I need to take a reading off an instrument at every point within this volume, with a displacement of 0.1mm between readings.

The lidar systems I've looked at are only good to within 1mm. Ultrasonic is right out.

any ideas?

 

[Hesch]

Suggestions:

Distance-measurement ±0.1mm with laser is possible, but I think it's very expensive and slow.

With a number of cameras you can aim at something within the cube. The lens will distort the picture, but if you aim exact at a point, the position ( direction ) of this point will not be distorted. By measuring the exact direction of the cameras having aimed, the position of the point can be calculated by triangulation. Use spindels with no backlash and encoders to measure the number of degrees the spindles have to turn. Say 4 cameras, with 3 spindles and 3 encoders each = 12 spindles + 12 encoders. Also I think that 3 - 4 PC's is needed to control the cameras and calculate pictures "real time". It's like "hawk-eye" in tennis. The system has to be calibrated.

 

[berkeman]

Hello,

I'm looking for some ideas as to how I can get the position of a beacon/actuator to within 0.1mm in a cube 10m x 10m x 10m.

I need to take a reading off an instrument at every point within this volume, with a displacement of 0.1mm between readings.

The lidar systems I've looked at are only good to within 1mm. Ultrasonic is right out.

any ideas?

How are you translating your device within that volume? Maybe just use 3 accurate linear displacement mechanisms -- kind of like long versions of calipers...

http://i.stack.imgur.com/AtTpr.gif

 

[terryphi]

How are you translating your device within that volume? Maybe just use 3 accurate linear displacement mechanisms -- kind of like long versions of calipers...

http://i.stack.imgur.com/AtTpr.gif

Well, I'd perfer it to be able to be positioned without any sort of physical connection to a wall or something like that.

 

[berkeman]

Well, I'd perfer it to be able to be positioned without any sort of physical connection to a wall or something like that.

Can you say what the measurement is? If it involves moving an RF transceiver, then any contact measurement technique like the one I suggested will need to involve a non-conducting mechanical structure. The caliper technique can still be used, albeit with less resolution (but still meeting your 0.1mm requirement.

You might consider some sort of ultrasonic clicking sonar distance measuring means... Mount 3 directional clicker/emitter antennas on your device and synchronize the received blips with the transmitted signals. With some experimentation and tuning you may be able to get your resolution. Each of the x, y, z clicks would be done separately at each position, and it may help to have some simple ultrasound-absorbing material on the walls of the room where you have this 10mx10mx10m apparatus set up.

 

[DaveC426913]

And what are you going to do with that petabyte of data??

(Amusingly, an array of terabyte drives big enough to hold a single one of your data samples would have dimensions that are within an order of magnitude of the volume itself: a metre on a side.)

 

[256bits]

And what are you going to do with that petabyte of data??

(Amusingly, an array of terabyte drives big enough to hold a single one of your data samples would have dimensions that are within an order of magnitude of the volume itself: a metre on a side.)

I was wondering the same thing.
Not only that, but will he live long enough, his kids, his grandkids, ... to see the final data point taken.
Assuming of course, a 1 sec seek and record for each sequential data point.
Might be able to speed that up somewhat though.

 

[Baluncore]

I need to take a reading off an instrument at every point within this volume, with a displacement of 0.1mm between readings.

Is there some physical object within that cube?
Do you want the surface shape information for that object within the volume?

 

[DaveC426913]

Not only that, but will he live long enough, his kids, his grandkids, ... to see the final data point taken.
Assuming of course, a 1 sec seek and record for each sequential data point.

I get a MUCH more pessimistic answer than you.

The question is more: will we still be definably human when the data set is complete?

10¹⁵ points, even at 100 samples per second, will take more than 300,000 years!