<?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>🏫 教学 on Elon&#39;s AD Insight</title>
    <link>https://auto-driving-blog.vercel.app/tags/-%E6%95%99%E5%AD%A6/</link>
    <description>Recent content in 🏫 教学 on Elon&#39;s AD Insight</description>
    <generator>Hugo</generator>
    <language>zh-cn</language>
    <lastBuildDate>Sun, 19 Jul 2026 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://auto-driving-blog.vercel.app/tags/-%E6%95%99%E5%AD%A6/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>论文精读｜Teaching VLA Models What to See and Where to Look：DriveTeach-VLA 的视觉教学三段式</title>
      <link>https://auto-driving-blog.vercel.app/posts/paper-reading/driveteach-vla%E7%B2%BE%E8%AF%BB/</link>
      <pubDate>Sun, 19 Jul 2026 00:00:00 +0000</pubDate>
      <guid>https://auto-driving-blog.vercel.app/posts/paper-reading/driveteach-vla%E7%B2%BE%E8%AF%BB/</guid>
      <description>DriveTeach-VLA 指出驾驶 VLA 的瓶颈不在语言推理而在视觉空间 grounding，核心问题不是&amp;rsquo;想不清&amp;rsquo;而是&amp;rsquo;看不对&amp;rsquo;。它提出三段式视觉教学管线——DVD 蒸馏教模型&amp;rsquo;看什么&amp;rsquo;、2D 轨迹引导提示教&amp;rsquo;往哪看&amp;rsquo;、TGP-guided GRPO 教&amp;rsquo;怎么动&amp;rsquo;——层层递进注入驾驶感知先验。在 Qwen2.5-VL 上刷新 NAVSIM 与 nuScenes 双榜单 SOTA。</description>
    </item>
  </channel>
</rss>
