APEX: Empowering LLMs with Physics-Based Task Planning for Real-time Insight

Wanjing Huang, Weixiang Yan, Zhen Zhang, Ambuj Singh

Keywords: Physics-Enhanced LLMs, Graph-Based Perception, Task Planning, Predictive Simulation

TL;DR: We build a framework that helps LLMs quantify when a cat will collide with them and assess the physical outcomes of different escape routes (e.g., whether they’ll crash into a table) by feeding them physics-based simulations.

Abstract:

Large Language Models (LLMs) demonstrate strong reasoning and task planning capabilities but remain fundamentally limited in physical interaction modeling. Existing approaches integrate perception via Vision-Language Models (VLMs) or adaptive decision-making through Reinforcement Learning (RL), but they fail to capture dynamic object interactions or require task-specific training, limiting their real-world applicability.
We introduce APEX (Anticipatory Physics-Enhanced Execution), a framework that equips LLMs with physics-driven foresight for real-time task planning. APEX constructs structured graphs to identify and model the most relevant dynamic interactions in the environment, providing LLMs with explicit physical state updates. Simultaneously, APEX provides low-latency forward simulations of physically feasible actions, allowing LLMs to select optimal strategies based on predictive outcomes rather than static observations.
We evaluate APEX on three benchmarks designed to assess perception, prediction, and decision-making: (1) Physics Reasoning Benchmark, testing causal inference and object motion prediction; (2) Tetris, evaluating whether physics-informed prediction enhances decision-making performance in long-horizon planning tasks; (3) Dynamic Obstacle Avoidance, assessing the immediate integration of perception and action feasibility analysis. APEX significantly outperforms standard LLMs and VLM-based models, demonstrating the necessity of explicit physics reasoning for bridging the gap between language-based intelligence and real-world task execution.

Primary Area: applications to robotics, autonomy, planning

Submission Number: 13665