GreenWAM: A World Foundation Model for Robotics
We present GreenWAM, a world foundation model for robotics, together with the recipe that produces it: (1) a shared anatomical action space that describes 20 robot families, from bimanual arms to dexterous humanoids, in common joint coordinates; (2) joint training on these families in visual, action-conditioned and world-action modes; and (3) policies that split an action expert off the generator.
ProofPaper ↗
Key points
- A robot needs a world model that predicts how its actions will change the scene and a policy that chooses those actions.
- We ask whether one robot world model can serve both roles across different bodies, and what makes this work.
- We further adapt GreenWAM to a real humanoid, where it succeeds more often than policies adapted from $π{0.5}$ and from Cosmos 3 Edge, and distill it into a single-step policy for real-time control.
- Controlled experiments yield three findings: robot specialization improves world prediction and control together; an action expert helps when split off at the start of adaptation; and keeping world-model modes during policy adaptation preserves robustness to scene changes.
Sources (2)
- [1]GreenWAM: A World Foundation Model for RoboticsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 13, 05:31 PM
We present GreenWAM, a world foundation model for robotics, together with the recipe that produces it: (1) a shared anatomical action space that describes 20 robot families, from bimanual arms to dexterous humanoids, in common joint coordinates; (2) joint training on these families in visual, action-conditioned and world-action modes; and (3) policies that split an action expert off the generator.
A robot needs a world model that predicts how its actions will change the scene and a policy that chooses those actions.
- [2]GreenWAM: A World Foundation Model for RoboticsHugging Face Daily Papers · Oct 13, 12:00 AM · same content
Extractive summary: sentences quoted from the sources.
Before this
- Oct 12, 2026VLA-GAUGE: Systematically Evaluating Generalization and Robustness in Vision-Language-Action Models
- Oct 12, 2026World Models Need a Physical Registry: Learning Internal Predictive States for Physically Plausible World Models
- Oct 12, 2026PINT: A Unified Model with Shared Predictive Intent for Robot Manipulation
- Oct 11, 2026$Δ$ Flow Matching for Robot Foundation Models
- Oct 11, 2026StreamingWAM: Streaming World Action Modeling via Chunk-Wise Staircase Flow Matching
- Oct 10, 2026Verify, Act, Distill: Turning Test-Time Compute into Test-Time Adaptation for Vision-Language-Action Models