W2A: Distilling World Priors into Action Models for Real-Time Robotic Control
cs.RO
Submitted: 2026-11-02
Updated: 2026-11-02
Code: https://github.com/nvidia-cosmos/cosmos-predict2
Project page: https://misaka0502.github.io/W2A
Terminology
Sources
- OpenVLA: An Open-Source Vision-Language-Action Model
- $\pi_0$: A Vision-Language-Action Flow Model for General Robot Control
- GR00T N1: An Open Foundation Model for Generalist Humanoid Robots
- ReconVLA: Reconstructive Vision-Language-Action Model as Effective Robot Perceiver
- Video Prediction Policy: A Generalist Robot Policy with Predictive Visual Representations
- mimic-video: Video-Action Models for Generalizable Robot Control Beyond VLAs
- Unified Video Action Model
- DiT4DiT: Jointly Modeling Video Dynamics and Actions for Generalizable Robot Control
- Fast-WAM: Do World Action Models Need Test-time Future Imagination?
- Motus: A Unified Latent Action World Model
- Cosmos Policy: Fine-Tuning Video Models for Visuomotor Control and Planning
- Scalable Diffusion Models with Transformers
- Flow Matching for Generative Modeling
- SmolVLA: A Vision-Language-Action Model for Affordable and Efficient Robotics
- Gemini Robotics: Bringing AI into the Physical World
- RDT-1B: a Diffusion Foundation Model for Bimanual Manipulation
- Galaxea Open-World Dataset and G0 Dual-System VLA Model
- ACoT-VLA: Action Chain-of-Thought for Vision-Language-Action Models
- A Pragmatic VLA Foundation Model
- ${\pi}_{0.7}$: a Steerable Generalist Robotic Foundation Model with Emergent Capabilities
Related papers
- Rethinking Actions: Continuity-Aware Priors for World-Model Planning
- MOSAIC: Modular Obstacle and Spatial-Aware Framework for Manipulation In Cluttered Scenes
- SAGAS: Semantic-Aware Graph-Assisted Stitching for Offline Temporal Logic Planning
- DexSeed: Expanding Human Hand-Object Interactions into Robot Demonstrations for Dexterous Policy Learning
- LaST0.5: A Generalist World Action Model with Interaction-Centric Latent Reasoning
- CIALLO: Learning Shared Intent Representations for Cross-Embodiment Manipulation