New project! Flow Policy Gradients for Robot Control
tldr; a simple online RL recipe for training and fine-tuning flow policies for robots
co-led w/ @redstone_hong: hongsukchoi.github.io/fpo-control
Brent Yi
@brenthyi
Joined November 2009
- tyro 1.0 is out ๐ฃ This has been a pet project/niche interest of mine for ~4 years now, so it's a bit of a sentimental moment... github.com/brentyi/tyro
- At NeurIPS until Sunday, looking forward to meeting/catching up with people ๐ Also poster today with @ruilong_li 11AM-2PM, Exhibit Hall CDE #4409!For everyone interested in precise ๐ทcamera control ๐ท in transformers [e.g., video / world model etc] Stop settling for Plรผcker raymaps -- use camera-aware relative PE in your attention layers, like RoPE (for LLMs) but for cameras! Paper & code: liruilong.cn/prope/
- New project from @kevin_zakka I've been using + helping with! Ridiculously easy setup, typed codebase, headless vis => happy ๐I'm super excited to announce mjlab today! mjlab = Isaac Lab's APIs + best-in-class MuJoCo physics + massively parallel GPU acceleration Built directly on MuJoCo Warp with the abstractions you love.
- Humanoid motion tracking performance is greatly determined by retargeting quality! Introducing ๐ข๐บ๐ป๐ถ๐ฅ๐ฒ๐๐ฎ๐ฟ๐ด๐ฒ๐๐ฏ, generating high-quality interaction-preserving data from human motions for learning complex humanoid skills with ๐บ๐ถ๐ป๐ถ๐บ๐ฎ๐น RL: - 5 rewards, - 4 DR
