Skip to content

microsoft/FlowDAgger

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

FlowDAgger

Reference implementation for the paper FlowDAgger: Human-in-the-Loop Adaptation of Generative Robot Policies in Latent Space.

Project page: https://microsoft.github.io/FlowDAgger

FlowDAgger is latent-space DAgger for flow-matching and diffusion based robot policies. Instead of fine-tuning the base policy, it learns a small steering network that predicts the initial noise fed to the policy's sampler. Expert corrections are mapped back into that noise space by inverting the policy's sampling ODE, and the steering network is trained on the inverted targets with a behavior-cloning loss.

This repo is a minimal, self-contained reference implementation on the pi0.5 base policy (JAX / openpi), running the MetaWorld assembly task.

How this example works

  1. Roll out the base policy. A steering network predicts the sampling noise.
  2. An intervention handler hands control to a scripted expert.
  3. Each expert action chunk is inverted through the policy's sampler to recover the noise that would have produced it.
  4. The steering network is trained to predict those noise targets (MSE).
  5. Repeat. Over time the steering network reproduces expert behavior without touching the base-policy weights.

Layout

shared/             scripted expert, task registry, intervention handler
flowdagger_pi05/    the experiment: JAX, pi0.5 base, MetaWorld assembly
                    (openpi is a git submodule under flowdagger_pi05/openpi)

Getting started

git submodule update --init flowdagger_pi05/openpi

Then follow flowdagger_pi05/README.md for install and the exact launch command. The pi0.5 checkpoint is fetched from the Hub automatically on the first run.

License

MIT. See LICENSE.

About

FlowDAgger: Human-in-the-Loop Adaptation of Generative Robot Policies in Latent Space

Resources

License

Code of conduct

Security policy

Stars

12 stars

Watchers

0 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages