You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This issue is discovered by @bm-hellorobot. What is seen by the webcam that looks for the tong Aruco markers can affect the performance of the system. Part of the Dex Teleop system is a Linux process (launched through the multiprocessing library) that looks for Aruco markers in the webcam image. This image processing process is nominally running at 30hz. However, if the webcam sees a background with a large number of features that could be Aruco markers (at least according to the preprocessing steps used in OpenCV's Aruco marker detection routine), such as square white popcorn tile on the ceiling, then Aruco processing slows way down and the overall rate of the process drops. Then, the performance of the system is affected. In his testing, Blaine saw it drop down to 6hz with the popcorn ceiling, but jump back up to 30 hz when the background was blocked out.
To reproduce this issue, use the goal_from_teleop script. This reports the frequency of its last ten points so you can see it change in real time as you move the camera.
The text was updated successfully, but these errors were encountered:
This issue is discovered by @bm-hellorobot. What is seen by the webcam that looks for the tong Aruco markers can affect the performance of the system. Part of the Dex Teleop system is a Linux process (launched through the multiprocessing library) that looks for Aruco markers in the webcam image. This image processing process is nominally running at 30hz. However, if the webcam sees a background with a large number of features that could be Aruco markers (at least according to the preprocessing steps used in OpenCV's Aruco marker detection routine), such as square white popcorn tile on the ceiling, then Aruco processing slows way down and the overall rate of the process drops. Then, the performance of the system is affected. In his testing, Blaine saw it drop down to 6hz with the popcorn ceiling, but jump back up to 30 hz when the background was blocked out.
To reproduce this issue, use the goal_from_teleop script. This reports the frequency of its last ten points so you can see it change in real time as you move the camera.
The text was updated successfully, but these errors were encountered: