SNU Visual Computing Lab shares research overview: teaching robot hands human-level dexterity
CSProfKGD · x · 2026-10-18
Prof. Hanbyul Joo's Visual Computing Lab at Seoul National University published a one-page overview of its research: building AI that understands and acts in the physical world, grounded in computer vision, ML, graphics and robotics.
The lab's focus is dexterous manipulation — taking the human body as the reference and training robot hands and humanoids directly from internet videos and in-lab motion capture. Selected work includes Co²Skill (whole-body control via skill composition), DeVI (physics-based human-object interaction via synthetic video imitation), ZeroDex (zero-shot long-horizon manipulation via multi-view 3D-grounded VLM reasoning), AutoDex (CoRL 2026, automated grasping data collection), and HRDexDB, a 4D dexterous grasping dataset across human and robot embodiments.
More from Embodied
- Ropedia Super Week: world model W0, dexterous bimanual controller DONIC and more — liuziwei7 · 2026-10-18
- Driver with nerve damage logs 102-day Tesla FSD streak, 95% autonomous — Vjeux · 2026-10-18
- Tesla owner with 26,000 km in one year explains why he won't buy FSD — michalmalewicz · 2026-10-18
- Robotics to demand 363 TWh of electricity globally by 2035, per Wood Mackenzie — ingliguori · 2026-10-18
- Mind Children's Codey: a child-sized companion humanoid for elders, pilots in 2027 — CyberRobooo · 2026-10-18
- SharpaDex v1.0: 300 hours, 28,993 bimanual dexterous demos with fingertip tactile sensing, open-sourced — DJiafei · 2026-10-18