Looking for ideas on how to track movement

[davidd31415]

I would like to track the movement of a part over a 1-2' arc as part of a durability test on components that are moved by motors. The motors turn about 90 degrees and the components travel in 1-2' arcs. The motors are actuated in extreme environments (-40C, 60C/80%RH, salt humidity, dust) for several days or weeks at a time.

I would like to produce graphs showing the speed of the moving component as a function of time so that any speed fluctuations that occur while the motor is running can be realized. Tracking the position with a resolution of an inch or better would be ideal.

Would RFID be useful in this application? I am curious what technology is used in the making of computer-animated films, where sensors are mounted to people and the movement is then reproduced on computer screens. I’ve considered machine vision but expect integrating it in extreme environments will be difficult.

Currently I have a system setup that monitors the extreme positions of the moving components using magnetic sensors and magnets mounted on the moving components. I’ve thought about mounting extra magnetic sensors but am hoping to come up with something better.

Any ideas will be appreciated!

Thanks,

Dave

 

[Skogsgurra]

Are the components attached to an arm that is attached to motor shaft? If so, an encoder, resolver, even a potentiometer (non-contacting) can do the job.

If not, then you need something else.

The human-to-computer thing is/was(?) often done using IR diodes and a 2D sensor that outputs two voltages proportional to X and Y position. The IR easily penetrates dirt, but the detector needs protective flow of air - and an occasional water spritz now and then.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[geekEE]

Are you trying to measure an angle or an X-Y position or something else? You mention "resolution of an inch or better" and "1-2' arcs". Do you mean 1-2 minutes or 1-2 feet?

If you are measuring an angle, can you just put a rotary encoder on the motor?

 

[davidd31415]

I'm trying to measure an X-Y position, 1 to 2 foot arcs.

I can not disassemble the parts to access the motor or shaft unfortunately. Mounting a rotary potentiometer on the outside of the part presents only mounting problems so I'll give that some more thought.

I researched the computer-animation motion tracking that I first asked about, "motion capture" seems to be the correct term for it. A magnetic motion capture system that can be purchased with only 1 sensor (instead of enough for body movement) designed for 2D looks like it might do the job well but I'm not sure where to find such a thing.

 

[ScottyUK]

Can the motion of the part be implied if the speed of the motor is known? If the motor has a solid slip-free mechanical drive to the rest of the assembly then the motion of the part can presumably be predicted from shaft speed. Measuring shaft speed is usually pretty simple.


----------------------------------

Sometimes I only open my mouth to swap feet...

 

[IRstuff]

how much budget? how much accuracy? how fast? how much room? how much line of sight available?

Without these answered, you cannot get a relevant solution.

TTFN

 

[IRstuff]

Sorry, missed the 1 inch resolution.

TTFN

 

[IRstuff]

see:

http://www.lyric.com/fcp-plugins/creating-tracks/creating-tracks.htm#tracking

http://www.perceptivu.com/TargetTrackingSoftware.html

The latter is a standalone product, while the former is a plug-in for MAC-based Final Cut Studio:

http://www.apple.com/finalcutstudio/finalcutpro/

Note that unless you want to spend lots of money, you're limited to 30 Hz frame rate for image-based systems, which might be insufficient for your application

TTFN

 

[davidd31415]

Regarding speed, 10Hz would be sufficient. Under $10k budget. There is line of sight but I am not sure how feasible it will be to use optical tracking in the extreme environments (the humidity/dust). Finding a CCD that operates in -40C and 60C/80%RH would be nice though; I'm sure something can be designed to blow off dust/wipe off humidity.

 

[IRstuff]

There's also ultrasonic:

http://www.equator.ac.uk/index.php/articles/670

http://www.vrdepot.com/vrteclg.htm

Cameras shouldn't be an issue:

http://www.cohu-cameras.com/pdf/3640.pdf

TTFN

 

[l3city]

"I would like to produce graphs showing the speed of the moving component as a function of time so that any speed fluctuations that occur while the motor is running can be realized."

If you scale it down, try to measure how much (in lenght) the motor shaft rotates and the time it takes to get to the desired position, knowing that you will be able to know your answer.

 

[VEBill]

"Would RFID be useful in this application?"

No.

"I am curious what technology is used in the making of computer-animated films, where sensors are mounted to people and the movement is then reproduced on computer screens."

What are mounted to the actors are reflective balls (not sensors). The balls are tracked in 3D using multiple IR cameras.

 

[davidd31415]

"What are mounted to the actors are reflective balls (not sensors). The balls are tracked in 3D using multiple IR cameras."

From what I've read that sounds more like optical motion capture...

http://www.metamotion.com/motion-capture/magnetic-motion-capture-1.htm

http://www.e-motek.com/simulation/about_us/mocap.html

http://www.wave-report.com/tutorials/MoTrak.htm

That last site claims line of sight is not required for magnetic. I found a reference to an article discussing speed and position tracking of a catheter (for autonomous insertion)- perhaps just a few sensors rather than a few dozen. Still looking for a place to start with this technology though.

 

[IRstuff]

Magnetic is very finicky and requires lots of calibration to account for alterations of the field distribution caused by magnetic materials within the field. Your motors under test would probably make magnetic sensors freak out.

TTFN

 

[waross]

I realise that you have said that the motor shaft not accesible, but:
When you consider the cost and dificulty of measuring speed accurately at very low RMPs, it is worth another try.
The resolution of your speed will be so much greater if you measure the motor speed that there is almost no comparison.
If you cannot find an encoder for the motor shaft, you may use a light and sensor to count fan blades passing.
A high quality encoder may give position indication as well as speed.
If a graph of relative operating speed is all that is required you may use a simple RPM indication. As a result of this, we are faced with several problems. Dirt and humidity, and attemting to accurately measure very low operating speeds.
It has been my experience that when one condition of a problem leads to such a magnification of difficulty, it is time to re-evaluate the original condition.
One of the stated conditions of the problem is the unavailability of the motor shaft.
It may be less difficult to modify the restricting condition than to work around it. In this case, try harder to access the motor shaft for measurement. If you are able, a lot of other difficulties will go away and the problem becomes very simple.
respectfully

 

[DaveScott]

Would a pull-wire encoder be useful?

 

[davidd31415]

Pull-wire; I have not worked with one before but may be useful if it pulls back (like a tape measure). I'll give that some thought; I'm taking a closer look at machine vision too.

I realize the motor shaft would be best for tracking motor movement but I trully need to track the position of the moving component; there are gears and clutches in between and accessing the motor shaft simply isn't an option.

 

[DaveScott]

Yes, you are right, they work pretty much like a tape measure (spring return), except you can get a very accurate pulse output. I think it might be possible that they give you much better performance than you require except for one thing: they can only measure how much wire is pulled out and not its direction. If your arc is not significant, then perhaps you could ignore the curve, but on the other hand you could work in some maths to compensate if necessary.
They are really easy to install, compared to a vision system!