2025/06/10 by Samuel Holt, Holt, Samuel, Max Ruiz Luyten +5 · 3 citations
Computer Science · Medicine · #Abstraction #Action (physics) #Adaptability #Artificial Intelligence in Healthcare and Education #Automated planning and scheduling #Key (lock) #Multimodal Machine Learning Applications #Plan (archaeology) #Quality (philosophy) #Topic Modeling
paper · pdf · doi:10.48550/arxiv.2506.09171
published in ArXiv.org
openalex publication_date 2025/06/10 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/01
Large Language Models (LLMs) are increasingly capable, but LLM agents still struggle to plan effectively in interactive, partially observable, long-horizon environments when search is unguided or recent history is insufficient. We introduce LWM-Planner, a fact-augmented lookahead planning framework that improves agent behavior purely through in-context learning. After each episode, the agent extracts task-critical atomic facts from its trajectories, validates candidates with a lightweight predictive-consistency filter (and optionally compresses them), and uses the resulting fact set to condition action proposal, single-step latent world-model simulation, and state-value estimation. Planning then proceeds via recursive, depth-limited lookahead over candidate trajectories conditioned on the accumulated facts and recent history, enabling online improvement without parameter updates. We provide abstraction-style motivation: treating facts as reducing state aliasing (proxy εsim) and fact-conditioned simulation as lowering one-step error (proxy δmodel), without claiming formal guarantees. Empirically, on text FrozenLake variants, CrafterMini, and ALFWorld, the approach improves cumulative return over ReAct/Reflexion and search-only baselines, suggesting that additional test-time search is most useful when grounded by compact, experience-derived facts.