R200 acquireframe() returns EXEC_ABORTED

[Susan_S_]

Sometimes, but not every time, my calls to PXCMSenseManager::acquireframe(true) returns a status of pxcmStatus.PXCM_STATUS_EXEC_ABORTED. 

This doesn't appear to be a documented return value for this function call - does anyone have any insight on what may be causing this return? 

[Xusheng_L_Intel]

I am investigating this issue now and will keep you post when I find the answer. Thanks!

[Susan_S_]

Any progress or response on this issue?  I continue to experience this undocumented return value and don't know how to resolve it.

Thanks,

Susan

[Jonas_F_1]

I am getting the same error. It fails randomly after a period of time.

Is it a faulty camera?

[Jonas_F_1]

Now it got even weirder. The time it would live before failure got shorter and shorter to about a second or two. And now it's stuck on Init Started.

[Xusheng_L_Intel]

Could you provide the log so we can do further investigation? You follow this doc https://software.intel.com/sites/landingpage/realsense/camera-sdk/v1.1/documentation/html/index.html?sdkinfo_logging_control.html to collect. Thanks!

Jonas F. wrote:

Now it got even weirder. The time it would live before failure got shorter and shorter to about a second or two. And now it's stuck on Init Started.

[Jonas_F_1]

I just tested the camera on a different computer and there it seems to work without any issues, so appears to be a software issue.

On the problematic system I had to uninstall the virtual driver for the camera to even start. I will get the log when back at the other computer.

[jb455]

Was the cause of this ever discovered? I've started getting this error message too, at seemingly random intervals: sometimes I can leave the camera streaming for 10 or 20 minutes before hitting it, and other times it occurs after 10 seconds of streaming. If it "gets into the habit" of throwing this error a short time after the stream initialises, the only thing I can do to make it last longer is restart the device (merely restarting the DCM service makes no difference).

It has only appeared since I removed some bottlenecks in my application, allowing the camera frame rate to increase (it was previously up to 10-15fps, now it's very smooth, at least 30). Maybe my application is now running /too/ fast, and the camera sometimes doesn't have a frame ready when I ask it for one? I should note that it works fine on my PC; we only get it when testing on a MIIX 700 or Spectre X2.

[jb455]

I've managed to get around this now.

From further testing, I realised that if I stepped past the error and continued with pSenseManager.QuerySample(), on the frame /after/ the EXEC_ABORTED status hits, one of the color or depth components of the sample was null (almost always the colour image).

I eventually found that setting the frame rate explicitly on the EnableStream() method to be the same for the colour and depth streams (previously I had left that argument blank, only specifying stream type, width and height) stops the error coming up on AcquireFrame completely.

My guess is that after some time, the colour and depth streams were coming out of sync so AcquireFrame could only get one of them and hence threw the error. If my guess is accurate, is that expected behaviour from the SDK, or does it represent some sort of bug?