Fast Pose Tracking of Rigid Objects with Compact Pose Graph Optimization

Name: Fast Pose Tracking of Rigid Objects with Compact Pose Graph Optimization
Uploaded: 2026-10-14T22:35:01.529200+00:00
Duration: 301 s
Description: The work proposes a lightweight tracking module that uses relative pose constraints instead of dense point correspondences for online 6D pose estimation. By weighting these constraints based on geometric alignment uncertainty, the method maintains temporal consistency and accuracy while running much faster than existing optimization-based trackers.

The work proposes a lightweight tracking module that uses relative pose constraints instead of dense point correspondences for online 6D pose estimation. By weighting these constraints based on geometric alignment uncertainty, the method maintains temporal consistency and accuracy while running much faster than existing optimization-based trackers.

Open the full episode