让大模型规划自动驾驶轨迹更准更可靠，避免幻觉和失控。
GAPL: Grounded Action-effect Policy Learning for LLM-Based Trajectory Planning
- 用大模型估算动作影响，再通过仿真验证并修正结果。
- 在四个高速场景中碰撞率、误差等指标全面优于基线。
- 适合做智能驾驶决策系统的研究者与开发者参考。
自动驾驶轨迹规划需兼顾高层推理与底层控制。大语言模型（LLMs）虽具备语义丰富的规划能力，但受限于幻觉推理、环境动态缺乏真实感以及控制精度不足。本文提出GAPL（Grounded Action-effect Policy Learning），一个统一框架，将基于大模型的效应估计、基于仿真的效应校准与策略优化整合为闭环系统。GAPL包含三个模块：(1) 基于大模型的效应评估器，用于结构化多维动作-效应估计；(2) 基于仿真的效应校准器，通过模拟推演预测符合动力学一致性的效应；(3) 效应感知决策器，利用教师模型将大模型的效应估计与仿真结果对齐，指导基于PPO的策略学习。在四个Highway-env场景的实验表明，GAPL持续优于基线，平均降低碰撞率0.76、平均位移误差（ADE）0.86、最终位移误差（FDE）2.00，平均奖励提升1.44。
原文摘要 · Abstract (English)
Trajectory planning for autonomous driving requires both high-level reasoning and precise low-level control. Large Language Models (LLMs) offer semantic-rich planning capabilities, however, their application is limited by hallucinated reasoning, poor grounding in environment dynamics, and limited numerical precision in control. We propose GAPL (Grounded Action-effect Policy Learning), a unified framework that integrates LLM-based effect estimation, simulation-based effect grounding, and policy optimization into a closed-loop system. GAPL consists of three modules: (1) an LLM-based Effect Evaluator for structured multi-dimensional action-effect estimation; (2) a Simulation-based Effect Grounder that predicts dynamics-consistent effects from simulator rollouts; and (3) an Effect-Aware Decision Maker that grounds LLM effect estimates against simulation via a distiller to guide Proximal Policy Optimization (PPO)-based policy learning. Experiments on four Highway-env scenarios demonstrate that GAPL consistently outperforms baselines, achieving average reductions of {0.76, 0.86, 2.00} in collision rate, average displacement error (ADE), and final displacement error (FDE), and an average reward gain of 1.44.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。