Slot-MPC: Goal-Conditioned Model Predictive Control with Object-Centric Representations

arXiv 2026 58.5 method

TLDR

Slot-MPC uses object-centric world models with gradient-based MPC for efficient action planning in simulated robotic tasks.

Reasoning

The paper presents a novel integration of object-centric representations with differentiable MPC, showing improved planning efficiency and task performance in simulation. However, it lacks real-world validation and the abstract does not detail limitations or comparisons to state-of-the-art object-centric methods.

Read-first score

Read-first score 58.5, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 43.

Recency 6%

100

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

Methodology quality 18%

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

Citation impact 18%

69.9

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.6992213

Topical relevance 29%

61.4

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=code,github

Citation velocity 12%

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

Field roles

FrontierBridgeMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 374.

Keyword Scores

world model

world dynamics prediction

generative world model

interactive world model

model-based reinforcement learning world model

video world model

world simulator

Deep Analysis

Innovations

Combining object-centric world models with gradient-based Model Predictive Control (MPC) for action planning
Using slot-based representations from vision encoders to learn an action-conditioned object-centric dynamics model
Gradient-based MPC that directly optimizes actions through a differentiable world model, improving computational efficiency over sampling-based methods

Methodology

Slot-MPC uses vision encoders to learn slot-based representations that encode individual objects in the scene. These structured representations are used to learn an action-conditioned object-centric dynamics model. At inference time, the differentiable world model enables gradient-based MPC to directly optimize actions, allowing the agent to adapt to unseen situations. The approach is evaluated in an offline setting with limited state-action coverage on simulated robotic manipulation tasks.

Key Results

Slot-MPC improves both task performance and planning efficiency compared to non-object-centric world model baselines. In the offline setting with limited state-action coverage, gradient-based MPC outperforms gradient-free, sampling-based MPC.

Limitations

Only evaluated on simulated robotic manipulation tasks, not on real-world environments
Offline setting with limited state-action coverage may not reflect performance in online or data-rich scenarios

Resources

Paper