V-Dreamer: Automating Robotic Simulation and Trajectory Synthesis via Video Generation Priors

arXiv 2026 42.5 method, system, application

TLDR

V-Dreamer automates generating simulation-ready 3D manipulation environments and executable robot trajectories from language, using video generation priors and visual-kinematic alignment, with sim-to-real success.

Reasoning

The paper presents a strong integrated pipeline from language to simulation-ready scenes and executable trajectories, with empirical validation including sim-to-real transfer. However, the abstract offers limited detail on baseline comparisons and the generalizability of the approach beyond tabletop manipulation with one robotic arm.

Read-first score

Read-first score 42.5, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 28.

Recency 8%

100

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

Topical relevance 42%

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 25%

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

Reproducibility 25%

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

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 683.

Keyword Scores

world simulator

video world model

generative world model

world model

world dynamics prediction

interactive world model

model-based reinforcement learning world model

Resources

Paper