DexhandsIndustry hub

DEXTEROUS MANIPULATION

The world of dexterous hands.

Compare hardware. Discover applications. Share what works. An open reference and community for dexterous manipulation.

Geometry-Aware Dexterity

UC Berkeley / Google Brain / CMU research team

Geometry-aware multi-task policy learning for in-hand manipulation of objects with varied shapes.

What you can explore

Explore how object point clouds help a policy transfer between geometries.

Before you use it

The documented stack uses old Gym, MuJoCo 1.50, Python and CUDA versions. Plan a separate environment or port; published benchmark results are not a current hardware compatibility test.

Original contributors

Wenlong Huang, Igor Mordatch, Pieter Abbeel and Deepak Pathak

License / access: Code: MIT · check object assets and dependencies

Original code & documentation

Research overview

Editorial reference · checked 2026-09-22

Start a discussion

Published posts are public. First contributions are reviewed. Only share material you have permission to disclose.

Report a post