ACT / Action Chunking with Transformers
tonyzhaozh
An imitation-learning implementation that predicts chunks of robot actions from demonstrations.
What you can explore
Explore training and simulation examples as a baseline for collecting and learning manipulation skills.
Before you use it
The original ALOHA setup uses grippers. Dexterous-hand deployment requires adapting the action space, data and controller rather than reusing a policy unchanged.
Original contributors
Tony Z. Zhao and ACT collaborators
License / access: Code: MIT · check external data, models and hardware terms separately
Keep exploring
LeRobot
Tools for recording datasets, training policies and exploring robot-learning workflows.
Diffusion Policy
A visuomotor policy-learning framework that generates action sequences through diffusion.
HumDex
A humanoid dexterous manipulation stack combining teleoperation, data collection and policy learning.
Editorial reference · checked 2026-09-22