IMWM: Intuition Models Complement World Models for Latent Planning

arXiv 2026 55.2 method

TLDR

IMWM pairs a world model with an intuition model from demonstrations to improve latent planning, outperforming world-model-only planners on four pixel-based tasks.

Reasoning

The paper clearly identifies a limitation of world models alone (search bottleneck) and proposes a practical solution with three lightweight components. Strengths include strong empirical results across multiple tasks and a clean ablation. Weaknesses are the limited task diversity (all simulated) and lack of real-world validation.

Read-first score

Read-first score 55.2, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 41.

Recency 6%

100

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

Citation impact 18%

91.9

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.91889467

Methodology quality 18%

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

Topical relevance 29%

58.6

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

Reproducibility 18%

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

Citation velocity 12%

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

Field roles

FoundationFrontierBridge

Rank sensitivity

Stability: volatile; rank range: 408.

Keyword Scores

world model

model-based reinforcement learning world model

world dynamics prediction

generative world model

interactive world model

world simulator

video world model

Deep Analysis

Innovations

Pairing an intuition model (trained from demonstrations) with a world model for latent planning to address search bottlenecks
Retrieval Initialization: initializing planner's action proposals from retrieved demonstrations
Hybrid Cost: combining intuition score with world-model rollout cost
Reliability Gate: adjusting trust in intuition per setting

Methodology

IMWM combines a learned latent world model with an intuition model trained from demonstrations. Three lightweight components are introduced: Retrieval Initialization seeds the planner with actions from a retrieved demonstration; Hybrid Cost merges the intuition score with the world-model rollout cost; and a Reliability Gate modulates how much the planner relies on intuition. The method is evaluated on four pixel-based goal-reaching tasks (Two-Room, Reacher, Push-T, OGBench-Cube) against a world-model-only planner, using success rate as the metric.

Key Results

IMWM achieves higher mean success than the world-model-only planner on all four tasks, with the largest gains on Two-Room (99.2%, +11.5 percentage points) and OGBench-Cube (94.7%, +28.5 percentage points).

Resources

Paper