Gravity Vector on T265
I am running 2.34.0 version of librealsense under Windows 10 (latest release) and the acceleration vector looks incorrect on the T265. It is correct on the D400 series of cameras. To see the problem, use the rs-pose-predict app in the librealsense distribution and modify the following line:
std::cout << "Predicted " << std::fixed << std::setprecision(3) << dt_s*1000 << "ms " <<
"Confidence: " << pose_data.tracker_confidence << " T: " <<
predicted_pose.translation.x << " " <<
predicted_pose.translation.y << " " <<
predicted_pose.translation.z << " (meters) \r";
and change "translation" to "acceleration". All 3 values are near zero on my T265 which is incorrect. If the camera is vertical, the y value should be close to 9.8 - earth's gravitational constant.
In poking around in the pose data, I can see the 9.8 value so this appears to be a structure alignment issue.
Bob Davies
-
T265 questions are now handled on the RealSense GitHub forum Bob. Please re-post your question there! Thanks, and apologies for the inconvenience.
https://github.com/IntelRealSense/librealsense/issues
Please sign in to leave a comment.
Comments
1 comment