D405 performance lost then magically restored with no change
Hunting a bit of a weird failure mode and would appreciate some input!
I've got 2 D405 cameras that we verified as working great in our lab environment. These then went to their working environment, which is outdoors during the days, and stored in a shed overnight (which is not temperature or moisture controlled). After a few days we found 1 of the D405 cameras lost precision + accuracy (as measured with a ground truth test we have), then after another week we found the 2nd D405 camera exhibited similar failure modes. We brought these cameras back to the lab for investigation, and they sat on a desk (at room temperature) for a few days before we got around to investigating.
When we went to investigate these 2 "bad" cameras with our ground truth test, we found that magically they were functioning great again. We didn't do anything to them other than store them indoors between the "bad" and "good" test results. We've verified that our ground truth test performs the same in our 2 locations (lab + working site), so I don't believe it is a measurement issue.
I know vibration + extreme temperatures can affect D405 calibration and we can restore good calibration using the Dynamic Calibration Tool. However, I didn't expect the camera could lose a calibration then restore itself. One theory we're working on is if this is moisture related (where being stored indoors resulted in drying out internal components).
Does this loss-then-self-restoration of calibration sound like a possibility for the D405s? If yes, what are possible causes?
-
Hi Toni Divic There was a past case where a RealSense camera stopped working correctly after being in storage for a couple of weeks and then recovered itself whilst being shipped by post to another location and worked immediately upon arrival. So there is precedent for your theory that a change of location might recover a camera whose electronics have been affected by the environment that it had been stored in.
A RealSense 400 Series camera cannot automatically restore its calibration if the health of the calibration reduces as the re-calibration has to be performed manually. So it sounds as though the calibration stored in the camera was likely always okay and another factor was affecting the image quality and accuracy.
-
I cannot find the original case link but there was not much information on it. The camera whose hardware was affected by a period of storage was simply shipped by post from India to Intel by the user and when Intel received and lab-tested the camera it was found to be working normally. A cause was not officially identified.
-
The D456 and D457 models have an IP65 rated casing that protects against entry into the camera by dust or by water jets from any direction, making them a good choice for outdoor applications.
For the D415, D435 and D455 models, the company autoVimation offers protective enclosures rated at IP67 (a higher level of dust / water protection).
https://www.autovimation.com/en/enclosures-en/chameleon-xs-realsense-en
There is not a pre-made enclosure available for the D405 model, so a custom one would have to be created. A CAD model file can be downloaded for D405 to aid in the creation of a 3D printing blueprint for such an enclosure by visiting the link below and clicking on the Depth Cameras link to download a zipped folder of RealSense 400 Series CAD files, including D405.
https://dev.intelrealsense.com/docs/stereo-depth-camera-d400#cad-files
The material chosen for the front window of the enclosure should be clear like in the example in the image below.
In regards to temperature, you want to ideally prevent the camera's internal operating temperature from exceeding 35 degrees C or falling below 0 degrees C.
When the camera is being stored in an inactive unpowered state then it can withstand temperatures as low as -40 degrees C for a short period of time, with 0 degrees recommended as the minimum for prolonged exposure.
Please sign in to leave a comment.
Comments
7 comments