https://auto-driving-blog.pages.dev/tags/%E6%9C%BA%E5%99%A8%E4%BA%BA%E5%9F%BA%E7%A1%80%E6%A8%A1%E5%9E%8B/ Recent content in 机器人基础模型 on Elon's AD Insight https://auto-driving-blog.pages.dev/images/share.png https://auto-driving-blog.pages.dev/images/share.png Hugo zh-cn Fri, 18 Sep 2026 00:00:00 +0000 https://auto-driving-blog.pages.dev/posts/paper-reading/rdt%E4%B8%8Erdt2%E7%B2%BE%E8%AF%BB/ Fri, 18 Sep 2026 00:00:00 +0000 https://auto-driving-blog.pages.dev/posts/paper-reading/rdt%E4%B8%8Erdt2%E7%B2%BE%E8%AF%BB/ 从 DDPM 到 Flow Matching，从 1.2B 到 7B+400M，清华刘松铭团队的 RDT 系列完整拆解：为什么扩散模型特别适合机器人动作生成？多模态动作分布怎么建模？统一动作空间如何跨机器人迁移？三阶段训练配方如何对齐语言与连续控制？