- First, check if there’s an issue with the camera’s intrinsic parameters. If the intrinsic error is too large, it can lead to this type of error. If the camera hasn’t been hit or bumped, the intrinsic parameters can be optimized by rectification.
If the camera has been hit by something or fallen to the ground, please contact Mech-Mind’s technical support for assistance.
- Investigate the accuracy of the robot (compare measured values with the distance displayed on the robot’s teach pendant after the robot has moved for a certain distance).
- If the robot’s accuracy issues are ruled out, it may be related to the internal offset of the camera. Typically, this problem arises when there were issues during previous calibration processes, and compensation parameters had to be recalculated.
For details on when to use recalculated compensation parameters, please see the docs.
- Check the zero points of each robot axis to troubleshoot robot accuracy issues.
- If the robot’s issues are ruled out, it’s likely still a camera offset problem. The methods described above can be used to address this camera offset issue.
In certain on-site situations, if the point cloud output in the Mech-Vision project, which converts depth maps to point clouds, becomes a planar point cloud (or shows a significant difference compared to the point cloud displayed in Mech-Eye Viewer), it’s also related to the camera offset. The methods described above can be used to address this issue.