HRDexDB
Seoul National University / HRDexDB team
Paired human and robot grasp recordings with visual, kinematic and three-dimensional annotations.
What you can explore
Explore comparable grasps across human and robotic embodiments, including tactile-enabled subsets.
Before you use it
Use the published availability fields and explicitly verified pairings. The train table is a catalog, not a benchmark split; mesh and tracking revisions must stay aligned.
Original contributors
Jongbin Lim, Taeyun Ha and HRDexDB collaborators
License / access: Dataset: CC BY-NC 4.0 · check toolkit and model permissions
Editorial reference · checked 2026-09-22