Unlocking Smarter Device Control: Foresighted Planning with a World Model-Driven Code Execution Approach

arXiv 25.5 2025 47.2 method, application

TLDR

Proposes FPWC, a world model-driven framework for foresighted planning in mobile device control, achieving 44.4% relative improvement in task success rate.

Reasoning

The paper presents a clear contribution with a novel world model-based planning approach and demonstrates strong empirical results in both simulated and real-world settings. However, the abstract lacks details on the world model's architecture and generalization, and the keyword 'world model' is central while other specified keywords are not directly addressed.

Read-first score

Read-first score 47.2, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 26.

Recency 8%

86.7

Uses a gentle age decay so recent papers surface without erasing older foundations. 2025

Methodology quality 25%

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

Reproducibility 25%

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

Topical relevance 42%

37.1

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

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 376.

Keyword Scores

world model

world dynamics prediction

model-based reinforcement learning world model

world simulator

interactive world model

generative world model

video world model

Deep Analysis

Innovations

Development of a task-oriented, refinable world model at the outset of the task to enhance global environmental understanding
Foresighted action generation through iterative planning within the world model
Execution of actions in the form of executable code
Prioritization of natural language understanding and structured reasoning for decision-making

Methodology

The FPWC framework first constructs a task-oriented, refinable world model at the start of the task. It then performs iterative planning within this model to generate foresighted actions, which are executed as code. The approach is evaluated in simulated environments and on real mobile devices, comparing against previous methods.

Key Results

The method achieves a 44.4% relative improvement in task success rate compared to the state-of-the-art in the simulated environment.

Resources

Paper