Back to Tools
RoboMimic
Step-by-step RoboMimic guide covering BC, DAgger, and MimicGen workflows. Includes hardware requirements, dataset formatting, and integration with OpenArm and Mobile ALOHA.
Imitation learning framework. BC, DAgger, HBC. MimicGen for synthetic data augmentation.
Overview
RoboMimic provides algorithms (BC, DAgger, HBC) and infrastructure for imitation learning. MimicGen extends it by generating 50K+ synthetic demos from ~200 human demos. Supports simulation and real robots. MIT license.
Official Links
- robomimic.github.io — Project site
- github.com/ARISE-Initiative/robomimic — Code
A dataset format is step one; the loop trains on it, then scores the policy on real hardware.
Collect real-world data → License training data → Score a policy →