Bi-DexHands
PKU-MARL
A collection of bimanual dexterous manipulation tasks for reinforcement-learning experiments.
What you can explore
Compare single-agent and multi-agent learning setups across coordinated two-hand tasks.
Before you use it
Uses legacy Isaac Gym; environments and assets need compatible versions. Benchmark scores do not demonstrate real-hardware transfer.
Original contributors
Bi-DexHands authors and repository contributors
License / access: Apache 2.0 · check downloaded asset terms
Editorial reference · checked 2026-09-22