FlowEval: How Much Adaptation Does an Embodied Foundation Model Need at Deployment?
cs.RO
Submitted: 2026-11-08
Updated: 2026-11-08
Project page: https://www.figure.ai/news/helix-02
Terminology
Sources
- OpenVLA: An Open-Source Vision-Language-Action Model
- $\pi_{0.5}$: a Vision-Language-Action Model with Open-World Generalization
- Open X-Embodiment: Robotic Learning Datasets and RT-X Models
- Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Success
- Gemini Robotics 1.5: Pushing the Frontier of Generalist Robots with Advanced Embodied Reasoning, Thinking, and Motion Transfer
- MolmoAct2: Action Reasoning Models for Real-world Deployment
- World Action Models are Zero-shot Policies
- Don't Stop Pretraining: Adapt Language Models to Domains and Tasks
- Flow Matching for Generative Modeling
- Understanding deep learning requires rethinking generalization
- On Calibration of Modern Neural Networks
- Scaling Laws for Neural Language Models
- A Reduction of Imitation Learning and Structured Prediction to No-Regret Online Learning
- Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware
- RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control
- Evaluating Real-World Robot Manipulation Policies in Simulation
- Neural Spline Flows
- When Can Conditional Flow Matching Replace Pointwise Negative Log-Likelihood?
- Flow Actor-Critic for Offline Reinforcement Learning
- LAP: Language-Action Pre-Training Enables Zero-shot Cross-Embodiment Transfer
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