<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>⚡ DiffGRPO on Elon&#39;s AD Insight</title>
    <link>https://auto-driving-blog.pages.dev/tags/-diffgrpo/</link>
    <description>Recent content in ⚡ DiffGRPO on Elon&#39;s AD Insight</description>
    <image>
      <title>Elon&#39;s AD Insight</title>
      <url>https://auto-driving-blog.pages.dev/images/share.png</url>
      <link>https://auto-driving-blog.pages.dev/images/share.png</link>
    </image>
    <generator>Hugo</generator>
    <language>zh-cn</language>
    <lastBuildDate>Sun, 19 Jul 2026 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://auto-driving-blog.pages.dev/tags/-diffgrpo/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>论文精读｜ReCogDrive：小米EV强化认知端到端自动驾驶框架</title>
      <link>https://auto-driving-blog.pages.dev/posts/paper-reading/recogdrive%E7%B2%BE%E8%AF%BB/</link>
      <pubDate>Sun, 19 Jul 2026 00:00:00 +0000</pubDate>
      <guid>https://auto-driving-blog.pages.dev/posts/paper-reading/recogdrive%E7%B2%BE%E8%AF%BB/</guid>
      <description>华科×小米EV 提出 ReCogDrive，将 VLM 认知推理（场景理解与决策分析）、扩散规划器（连续轨迹生成）和 DiffGRPO 强化学习（安全优化）三阶段统一。它的层级化数据流水线自动生成含推理链的 VQA 标注，解决了 VLM&amp;rsquo;会想但不会开&amp;rsquo;的模态错配问题。在 NAVSIM 和 Bench2Drive 上达到 SOTA，验证了&amp;rsquo;认知-规划-优化&amp;rsquo;三分天下的 VLA 设计范式。</description>
    </item>
  </channel>
</rss>
