Grounded-Dreamer: Robot Learning from World Model Synthetic Data with Physical Grounding

arXiv 2026 34.6 method, system, application

TLDR

Grounded-Dreamer uses video world models and physics simulation to synthesize physically grounded robot manipulation data, training real-world policies without human demonstrations.

Reasoning

The paper offers a compelling combination of generative video realism and physical simulation for scalable robot data generation, with a clear filtering step for physically plausible rollouts and real-world validation. However, the abstract is truncated and lacks explicit comparisons or quantitative results, making it hard to assess robustness and generalizability.

Read-first score

Read-first score 34.6, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 47.

Recency 5%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Topical relevance 25%

67.1

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Methodology quality 15%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=result

Reproducibility 15%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Graph prestige 15%

5.7

PageRank-style prestige over the local citation graph built from available OpenAlex references. local_pagerank=0.0016

Citation impact 15%

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. cited_by_count=0

Citation velocity 10%

Citation velocity estimates citations per publication-year to reduce old-paper bias. velocity=0.00

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 370.

Keyword Scores

world model

video world model

generative world model

world simulator

world dynamics prediction

interactive world model

model-based reinforcement learning world model

Deep Analysis

Innovations

Introduces Grounded-Dreamer, a framework for synthesizing physically grounded robot manipulation data with video world models while avoiding human teleoperation demonstrations.
Combines generative video models with simulator physics engines to achieve both visual realism and physical validity.
Automatically reconstructs task-relevant objects and camera geometry from a small set of real robot task initial-state images and aligns them in simulation to collect physically valid trajectories.
Uses simulator trajectories to post-train both a video world model and an inverse dynamics model, enabling generation of visually realistic robot manipulation videos and corresponding robot actions from real images.
Filters physically implausible generated rollouts by replaying them in the reconstructed simulation and retaining only those that successfully complete the tasks.
Trains real-world robot policies solely on grounded synthetic trajectories without any human demonstrations or teleoperation effort.

Methodology

Starting from a small set of real images of robot task initial states, Grounded-Dreamer reconstructs task-relevant objects and camera geometry and aligns them in a physics simulator to collect physically valid trajectories. These trajectories post-train a video world model and an inverse dynamics model, allowing generation of visually realistic manipulation videos and corresponding actions from real images. Generated rollouts are replayed in the reconstructed simulation to filter for successful task completion, and real-world policies are trained only on the resulting grounded synthetic trajectories.

Key Results

Empirical results in real-world tasks demonstrate that Grounded-Dreamer can synthesize physically grounded data and substantially facilitate policy learning without requiring human teleoperation effort.

Resources

Paper