WEAVER, Better, Faster, Longer: An Effective World Model for Robotic Manipulation

arXiv 2026 56.4 method

TLDR

WEAVER is a multi-view world model for robotic manipulation achieving high fidelity, consistency, and efficiency, demonstrated on real hardware for policy evaluation, improvement, and planning.

Reasoning

The paper presents a novel world model architecture that simultaneously achieves fidelity, consistency, and efficiency, with strong empirical results on real robotic hardware. Its strengths include comprehensive evaluation across policy evaluation, improvement, and planning, while potential limitations may include reliance on specific model architecture choices not fully detailed in the abstract.

Read-first score

Read-first score 56.4, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 56.

Recency 5%

100

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

Citation impact 15%

87.8

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.87836959

Topical relevance 25%

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=baseline,evaluation,result

Reproducibility 15%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=code,github

Graph prestige 15%

5.7

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

Citation velocity 10%

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

Field roles

FoundationFrontierBridgeMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 432.

Keyword Scores

world model

world simulator

world dynamics prediction

generative world model

model-based reinforcement learning world model

interactive world model

video world model

Deep Analysis

Innovations

Simultaneously achieves fidelity, consistency, and efficiency in world models for robotic manipulation
Multi-view world model architecture trained with flow-matching loss to predict future latents and reward values
Key design decisions for model architecture, memory, and prediction objectives enabling long-horizon dynamic manipulation tasks

Methodology

WEAVER is a multi-view world model trained to predict future latents and reward values using a flow-matching loss. The architecture incorporates design choices across model structure, memory mechanisms, and prediction objectives to handle long-horizon dynamic manipulation tasks. Evaluation is performed on robotic hardware with baselines including the π0.5 robot foundation model and prior world models.

Key Results

WEAVER achieves a 0.870 correlation with real-world success rate for policy evaluation, a 38% real-world success rate improvement over the π0.5 foundation model for policy improvement, and a 14% improvement with 5-10× speedup over prior world models for test-time planning, along with better out-of-distribution performance.

Resources

Paper