<?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>轨迹评分 on Elon&#39;s AD Insight</title>
    <link>https://auto-driving-blog.vercel.app/tags/%E8%BD%A8%E8%BF%B9%E8%AF%84%E5%88%86/</link>
    <description>Recent content in 轨迹评分 on Elon&#39;s AD Insight</description>
    <image>
      <title>Elon&#39;s AD Insight</title>
      <url>https://auto-driving-blog.vercel.app/images/share.png</url>
      <link>https://auto-driving-blog.vercel.app/images/share.png</link>
    </image>
    <generator>Hugo</generator>
    <language>zh-cn</language>
    <lastBuildDate>Mon, 20 Jul 2026 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://auto-driving-blog.vercel.app/tags/%E8%BD%A8%E8%BF%B9%E8%AF%84%E5%88%86/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>论文精读：GTRS — 通用轨迹评分（CVPR25 自动驾驶挑战赛冠军）</title>
      <link>https://auto-driving-blog.vercel.app/posts/paper-reading/%E8%AE%BA%E6%96%87%E7%B2%BE%E8%AF%BB-2506-06664/</link>
      <pubDate>Mon, 20 Jul 2026 00:00:00 +0000</pubDate>
      <guid>https://auto-driving-blog.vercel.app/posts/paper-reading/%E8%AE%BA%E6%96%87%E7%B2%BE%E8%AF%BB-2506-06664/</guid>
      <description>GTRS（NVIDIA, CVPR 2025 自动驾驶挑战赛 E2E 赛道冠军）指出：现有轨迹评分器要么打「静态大词表」要么打「动态小候选」，两者都泛化差。GTRS 用三根支柱统一二者：(1) 扩散策略生成细粒度动态候选；(2) 在 16384 的超密集词表上训练评分器 + 词表 dropout，使推理时能在更小子集上强泛化；(3) 传感器旋转增强 + top-k 细化自蒸馏，提升域外鲁棒性。Navhard 榜 EPDMS 最高 49.4，逼近依赖真值感知的 PDM-Closed。</description>
    </item>
  </channel>
</rss>
