arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2603.10441cs.RO
收藏

KnowDiffuser: 一种基于知识的扩散规划器与大语言模型推理

KnowDiffuser: A Knowledge-Guided Diffusion Planner with LLM Reasoning

Monash University Malaysia(莫纳什大学马来西亚校区);University of Utah(犹他大学)

Fan Ding, Xuewen Luo, Fengze Yang, Bo Yu, HwaHui Tew, Ganesh Krishnasamy, Junn Yong Loo

AI总结:

本文提出KnowDiffuser,结合大语言模型的语义理解和扩散模型的生成能力,有效解决自动驾驶中语义到物理的鸿沟问题,实验表明其在开放环和闭环评估中表现优异。

摘要:

近年来,语言模型(LMs)在语义推理方面取得显著进展,使其能够应用于自动驾驶(AD)的高层决策。然而,LMs操作于离散的token空间,缺乏生成连续、物理上可行轨迹的能力。同时,扩散模型在生成可靠且动态一致的轨迹方面表现良好,但往往缺乏语义可解释性和与场景级理解的对齐。为了解决这些限制,我们提出了KnowDiffuser,一种结合语言模型语义理解和扩散模型生成能力的知识引导运动规划框架。该框架利用语言模型从结构化场景表示中推断出上下文感知的元动作,然后将其映射到先验轨迹,这些轨迹锚定了后续去噪过程。一个两阶段截断去噪机制高效地优化这些轨迹,同时保持语义对齐和物理可行性。在nuPlan基准测试中,KnowDiffuser在开放环和闭环评估中均显著优于现有规划器,建立了一个稳健且可解释的框架,有效弥合了自动驾驶系统中语义到物理的鸿沟。

英文摘要:

Recent advancements in Language Models (LMs) have demonstrated strong semantic reasoning capabilities, enabling their application in high-level decision-making for autonomous driving (AD). However, LMs operate over discrete token spaces and lack the ability to generate continuous, physically feasible trajectories required for motion planning. Meanwhile, diffusion models have proven effective at generating reliable and dynamically consistent trajectories, but often lack semantic interpretability and alignment with scene-level understanding. To address these limitations, we propose \textbf{KnowDiffuser}, a knowledge-guided motion planning framework that tightly integrates the semantic understanding of language models with the generative power of diffusion models. The framework employs a language model to infer context-aware meta-actions from structured scene representations, which are then mapped to prior trajectories that anchor the subsequent denoising process. A two-stage truncated denoising mechanism refines these trajectories efficiently, preserving both semantic alignment and physical feasibility. Experiments on the nuPlan benchmark demonstrate that KnowDiffuser significantly outperforms existing planners in both open-loop and closed-loop evaluations, establishing a robust and interpretable framework that effectively bridges the semantic-to-physical gap in AD systems.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑