RoboWorld: Fast and Reliable Neural Simulators for Generalist Robot Policy Evaluation

arXiv 2026 41.5 method, system, application

TLDR

RoboWorld uses a fast video world model with Step Forcing and VLM scoring to evaluate robot policies, achieving high correlation with real-world tests.

Reasoning

The paper introduces a novel evaluation pipeline with Step Forcing to reduce autoregressive mismatch, showing strong empirical correlation (Pearson's r=0.989) with real-world robot evaluation. However, the abstract lacks details on limitations, such as generalization across diverse tasks or failure modes of the world model.

Read-first score

Read-first score 41.5, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 47.

Recency 6%

100

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

Topical relevance 29%

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 18%

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

Reproducibility 18%

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

Citation impact 18%

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

Citation velocity 12%

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

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 319.

Keyword Scores

video world model

world model

world dynamics prediction

generative world model

world simulator

interactive world model

model-based reinforcement learning world model

Deep Analysis

Innovations

Automated evaluation pipeline RoboWorld that pairs a fast autoregressive video world model with task-progress-aware vision-language model scoring
Step Forcing technique combining anchored and one-step self-forwarded contexts to reduce train-test mismatch in autoregressive rollouts

Methodology

RoboWorld uses a fast autoregressive video world model to generate rollouts and a vision-language model to score task progress. Step Forcing is introduced to maintain action-observation dynamics while reducing mismatch between training and autoregressive inference. The pipeline's evaluation scores are correlated with real-world robot evaluation metrics.

Key Results

RoboWorld achieves strong alignment with real-world robot evaluation, with Pearson's r = 0.989 and Spearman's ρ = 0.970.

Resources

Paper