Reproduce “ArtReg: Visuo-Tactile based Pose Tracking and Manipulation of Unseen Articulated Objects” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Prajval Kumar Murali, Mohsen Kaboli
Researchers from BMW Group and academic institutions developed ArtReg, a framework enabling robots to track and manipulate unknown articulated objects without prior models by integrating visuo-tactile sensing through a Manifold UKF. This system achieved high accuracy in pose tracking and goal-driven manipulation while interactively detecting articulation types.
Key insights
- The synergistic integration of visuo-tactile sensing significantly enhances the robustness of object perception and manipulation, enabling robots to overcome individual modality limitations in challenging real-world conditions.
- Active, interactive manipulation strategies (pushing, holding-and-pulling) can effectively infer latent kinematic properties, such as joint types and limits, of previously unseen articulated objects.
- Utilizing a Manifold UKF on the SE(3) Lie Group provides a mathematically sound and robust approach for non-linear state estimation from partial and dynamic visuo-tactile data, outperforming traditional methods.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.