FlowDAgger: Online Imitation Learning for Robots
philfung · x · 2026-07-14
This repost introduces an online imitation learning method for robot manipulation called FlowDAgger. The author emphasizes that DAgger-style online learning is harder in practice than anticipated, and this work takes a step toward making it "truly usable" in real-world scenarios.
The post explains the background: modern physical AI models (including VLA and World-Action Models) typically use flow-matching to convert robot observations and noise vectors into actions. They usually start with offline imitation learning, followed by online imitation learning to "polish" the model via human corrections. This online correction is critical, as it determines whether the model is just "barely functional" or actually works in specific domains.
Three resources are included:
- Project page
- Paper
- Code
The original post also stresses that the performance difference becomes highly apparent when training on "your own robot and your own tasks."
More from Embodied
- Tesla expands Robotaxi rides to seven areas, including new Orlando and Tampa zones — elonmusk · 2026-07-22
- Hands-on robotics workshop on Saturday may be the last in-person session before August — StewartalsopIII · 2026-07-22
- NVIDIA pitches World Foundation Models as a way to scale physical AI data generation — MonaJalal_ · 2026-07-22
- RoboMME Podcast Preview: Benchmarking Memory for Robotic Policies — chris_j_paxton · 2026-07-21
- Gritt says an 8-person crew now installs 3,000 to 4,000 solar panels a day — HaktanSuren · 2026-07-21
- A helium-powered flying robot whale aims to be a quiet companion pet — chris_j_paxton · 2026-07-21