Awesome World Model Hub Papers · Datasets · Projects
← Back to papers

Unifying Object-Centric World Models and Diffusion Policy: A Hierarchical Framework for Multi-Stage Robotic Tasks

arXiv 2026 53.6 method, application

TLDR

A hierarchical framework combining object-centric world models with diffusion policy for multi-stage robotic manipulation tasks.

Reasoning

The paper introduces WorldDP, which uses a high-level world model for subgoal planning and a low-level diffusion policy for execution, showing strong empirical results on robotics benchmarks. Its strengths include a novel hierarchical design and object-centric representations, but limitations may include reliance on specific task domains and lack of theoretical guarantees.

Read-first score

Read-first score 53.6, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 35.

Recency 6%
100

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

Citation impact 18%
96.9

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.96940507

Methodology quality 18%
60

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

Topical relevance 29%
50

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%
30

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

Citation velocity 12%
0

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

Field roles

FoundationFrontierBridge

Rank sensitivity

Stability: volatile; rank range: 459.

Keyword Scores

world model
10
world dynamics prediction
8
world simulator
5
interactive world model
4
generative world model
3
model-based reinforcement learning world model
3
video world model
2

Deep Analysis

Innovations

  • Hierarchical framework combining object-centric world models with diffusion policy for multi-stage robotic manipulation
  • High-level world model used as a transition function to optimize feasible subgoals during runtime
  • Object-centric representations that decouple environmental entities for sequential planning per entity

Methodology

WorldDP employs a hierarchical approach with a high-level world model acting as a transition function to optimize feasible subgoals during runtime, which are then executed by a low-level Diffusion Policy. Object-centric representations are incorporated to decouple environmental entities, enabling sequential planning with respect to each entity. The framework is evaluated across several robotics benchmarks against existing baselines.

Key Results

WorldDP consistently outperforms existing baselines across several robotics benchmarks, demonstrating that coupling the world model's physically grounded planning with diffusion policy's efficient execution yields superior multi-stage performance.

Tags

world modelsdiffusion policyhierarchical frameworkmulti-stage manipulationobject-centricROAI