none
Kinect v2 operating temperature range RRS feed

  • Question

  • Hi there,

    I'm looking for any information on the safe operating temperature range for Kinect v2. I did find an article here:

    https://msdn.microsoft.com/en-us/library/dn782025.aspx?f=255&MSPPError=-2147217396

    but the link on that page to the temperature and humidity specifications does not appear to be publicly accessible.

    We have built a device with the Kinect v2 sensor that requires government certification and part of that certification process requires that we specify the valid operating temperature range for the device. This device is then tested at these extremes to ensure it still operates correctly. We've tried testing at 40 degrees to -5 degrees (Celsius) and it failed at both extremes. The Kinect was still working, but did not return depth measurements at the same accuracy it does when at room temperature.

    Thanks for any assistance you can provide.

    Paul

    Saturday, January 9, 2016 12:49 AM

All replies

  • The v2 (and most other depth sensors) is pretty temperature dependant. If you measure a depth point from when you cold start it, it will drift by about 20 mm over 45 minutes or so until the temperature/fan speed stabilises. After this time (so long as you dont shut off the sensor/lasers it keeps its temp and precision, (st dev < 0.5 mm for fusion tracking of a static object for 90 minutes).

    I read somewhere (I can try and find refs if needed) that the pulse shape changes slightly with temp, and this causes the depth measurement to be off. I believe (but this is not verified) that the pulse shape is calibrated at the warm shape, so the unit has an active heating element and fan control to attempt to keep it at this temp. Ive measured the temp of the diffusing elements in front of the lasers to be around 65C.

    I suppose at 40C the fans are not powerful enough to cool it against the background, and at -5 the heating element isnt powerful enough to get it up to idle temp. 

    If you dont mind, what use case are you expecting this kind of temperature range? Sounds interesting! If you need it over this range I would suggest building some thermal housing for the unit with large fans. 

     

    Monday, January 11, 2016 7:47 AM
  • Hi Phil,

    Thanks for the response. I gather from your answer that you've done some testing on this subject.

    We're building a solution that uses the kinect to measure the physical dimensions of objects. So far the results have been quite promising, as the solution is capable of measuring to accuracy we require (5mm). All of our testing has been at or around room temperature.

    As I mentioned in my original post, our solution must be certified by a government body in order for us to be able to actually bring it to market. The certification ensures that the measurements emitted by our solution have been tested to be correct to the accuracy specified.

    In order for our solution to be certifiable, it must be capable of producing accurate measurements within an operating temperature range of at least 30 deg C. By default the testing is done with a range of 40C to -10C. As mentioned above, the measurement accuracy was adversely affected at both of these temperatures. The effect was enough to push the measurements off by 1 deviation (5mm in this case). 

    So essentially I have to find a 30C window in which the device can perform well enough to achieve the accuracy we need. I would suppose (with the limited data I have) that 5C to 35C would be the best range to strive for, but I'm unsure of how well the device would perform at 5C.

    It's interesting to hear that the device has a heating element - do you know if there is a way to determine what the internal temperature is at any given time? It would be acceptable if we could detect that the unit wasn't warm enough and prevent operation until it reached a minimum safe temperature. If necessary we may have to add an external sensor to our solution to get this right, but obviously it would be far simpler to piggyback off of the device internals if possible.

    Do you have any thoughts on this? I'd appreciate any information you'd be willing to share...

    Thanks in advance,

    Paul



    • Edited by pdriegen Monday, January 11, 2016 4:15 PM
    Monday, January 11, 2016 4:14 PM
  • Hi Paul

    Do you have any knew informations about the safe operating temperatures for the Kinect v2?


    Wednesday, December 20, 2017 6:50 AM