Can the algorithm achieve sub-millimeter absolute positional accuracy?Does the accuracy of the algorithm depend on the accuracy of the camera, or can the algorithm improve the accuracy of the camera?

[sgg-al]

What is the absolute accuracy of the pose obtained by this algorithm? Has the author team done any relevant work to verify it?

[wenbowen123]

The accuracy - measured by the absolute error - is depending lots of things, such as image resolution, depth quality, camera intrinsics, etc. For reference, you can check the BOP benchmark, which can provide some sense on the relative accuracy among methods. We also have a recent work that uses FoundationPose for high-precision robotic industrial assembly.

[Cheney-36]

By the way, when will this work be integrated into the ISSAC lab

[wenbowen123]

@Cheney-36 it has been integrated into Isaac ROS. See our readme.

[Cheney-36]

@Cheney-36它已集成到 Isaac ROS 中。请参阅我们的自述文件。

I didn't express myself clearly. I am very curious about your AutoMate work, which is excellent. I saw plans to integrate it into Isaac Lab. Do you know the timeline?

[wenbowen123]

@Cheney-36 it has been released here.

[Cheney-36]

Thank you for your reminder, excellent work