{"id":10224,"date":"2026-07-15T20:02:14","date_gmt":"2026-07-15T12:02:14","guid":{"rendered":"https:\/\/infernews.com\/blog\/post-9a3c27b4\/"},"modified":"2026-07-15T20:02:14","modified_gmt":"2026-07-15T12:02:14","slug":"post-9a3c27b4","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/post-9a3c27b4\/","title":{"rendered":"[\u6280\u8853\u6587\u7ae0] Xiaomi-Robotics-U0 \u5c0f\u7c73\u7528\u4e16\u754c\u6a21\u578b\u6253\u901a\u6a5f\u68b0\u4eba"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/07\/pasted-2f8d7aa5f8cf.jpg\" alt=\"smileybones small\"><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">\u7576\u4e16\u754c\u6a21\u578b\u5f0f\u7684\u5f71\u50cf\u8207\u5f71\u7247\u751f\u6210\u80fd\u529b\u8981\u843d\u5230\u6a5f\u68b0\u4eba\u5834\u666f\uff0c\u96e3\u984c\u5514\u6b62\u4fc2\u51fa\u5716\u6216\u51fa\u7247\uff0c\u800c\u4fc2\u540c\u4e00\u500b\u5834\u666f\u55ba\u591a\u500b\u8996\u89d2\u4e0b\u90fd\u8981\u5408\u7406\uff0c\u7269\u4ef6\u5e7e\u4f55\u8981\u4e00\u81f4\uff0c\u4ef2\u8981\u7b26\u5408\u5514\u540c robot embodiment \u5605\u64cd\u4f5c\u9650\u5236\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Xiaomi-Robotics-U0 \u5c6c\u65bc world foundation model \u8def\u7dda\uff0c\u91dd\u5c0d\u7684\u6b63\u662f\u9019\u985e embodied synthesis \u5de5\u4f5c\uff1a\u4e00\u908a\u4fdd\u7559\u5927\u578b image and video generation model \u5df2\u5b78\u5230\u7684\u8996\u89ba\u77e5\u8b58\uff0c\u4e00\u908a\u88dc\u4e0a\u6a5f\u68b0\u4eba\u8cc7\u6599\u9700\u8981\u7684\u53ef\u63a7\u6027\u8207\u4e00\u81f4\u6027\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u5e38\u898b\u505a\u6cd5\u901a\u5e38\u4fc2\u7528\u6709\u9650\u7684\u6a5f\u68b0\u4eba\u8cc7\u6599\u53bb\u5fae\u8abf foundation model\uff0c\u4f46\u4f5c\u8005\u8a8d\u70ba\u5462\u7a2e\u7bc4\u5f0f\u5bb9\u6613\u72a7\u7272\u5927\u898f\u6a21\u9810\u8a13\u7df4\u5e36\u4f86\u7684\u6cdb\u5316\u80fd\u529b\u3002Xiaomi-Robotics-U0 \u6539\u7528 unified embodied synthesis \u8a2d\u8a08\uff0c\u628a text-to-image generation\u3001image editing\u3001embodied scene generation\u3001embodied transfer \u540c embodied video generation \u653e\u5165\u540c\u4e00\u500b 38-billion-parameter multimodal autoregressive model \u806f\u5408\u512a\u5316\uff0c\u5c07 embodied generation \u8996\u70ba foundation image and video generation \u7684\u5ef6\u4f38\uff0c\u800c\u5514\u4fc2\u53e6\u4e00\u689d\u5272\u88c2\u7684\u4efb\u52d9\u7dda\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u5462\u500b\u9805\u76ee\u6700\u6709\u7528\u7684\u5730\u65b9\uff0c\u5728\u65bc\u5b83\u4e0d\u53ea\u751f\u6210\u597d\u770b\u7684\u8cc7\u6599\uff0c\u800c\u662f\u751f\u6210\u53ef\u62ff\u4f86\u652f\u63f4\u6a5f\u68b0\u4eba\u5b78\u7fd2\u7684\u8cc7\u6599\u3002\u6587\u4e2d\u63d0\u5230\u5b83\u9996\u6b21\u652f\u63f4\u8de8\u591a\u7a2e robot embodiments \u7684\u9ad8\u54c1\u8cea multi-view scene generation\uff0c\u4ea6\u52a0\u5165 structured\u3001controllable embodied transfer\uff0c\u505a\u7d30\u7dfb\u7de8\u8f2f\u6642\u4ecd\u53ef\u4fdd\u7559 multi-view consistency \u540c interaction dynamics\uff0c\u5c0d\u8981\u505a\u6a21\u64ec\u8cc7\u6599\u64f4\u589e\u3001\u5834\u666f\u6539\u5beb\u3001\u64cd\u4f5c\u8ecc\u8de1\u914d\u5957\u751f\u6210\u7684\u5de5\u4f5c\u6d41\u5e6b\u52a9\u8f03\u5927\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u806f\u5408\u8655\u7406\u591a\u7a2e\u4efb\u52d9\uff0c\u6e1b\u5c11\u6bcf\u500b\u5834\u666f\u5404\u81ea\u505a\u6a21\u578b\u9069\u914d\u7684\u5272\u88c2\u6d41\u7a0b<\/li>\n<li>\u6838\u5fc3\u5dee\u7570\u5728\u65bc\u4fdd\u4f4f pre-trained world foundation model \u7684\u6cdb\u5316\uff0c\u540c\u6642\u52a0\u5165 embodied constraints<\/li>\n<li>\u652f\u63f4 multi-view scene generation \u8207 embodied transfer\uff0c\u91cd\u9ede\u653e\u5728\u5e7e\u4f55\u4e00\u81f4\u6027\u8207\u4e92\u52d5\u9023\u8cab\u6027<\/li>\n<li>\u751f\u6210\u7d50\u679c\u53ef\u4f5c\u70ba scalable data engine\uff0c\u670d\u52d9\u5f8c\u7e8c policy training<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">\u6548\u80fd\u4e0a\uff0cXiaomi-Robotics-U0 \u5728 single-step \u8207 sequential generation \u4efb\u52d9\u90fd\u505a\u5230 state-of-the-art\uff0chuman evaluations \u4e2d\u65bc embodied scene generation \u540c transfer \u8d85\u904e GPT-Image-2.0\uff0cembodied video generation \u5728 World Arena \u6392\u540d\u7b2c\u4e00\u3002\u66f4\u5be6\u969b\u7684\u6307\u6a19\u4f86\u81ea\u771f\u5be6\u64cd\u4f5c\u4efb\u52d9\uff1a\u5b83\u628a \u03c0 0.5 \\pi_{0.5} \u5728 out-of-distribution \u60c5\u6cc1\u4e0b\u7684 success rate \u7531 36.9% \u63d0\u5347\u5230 63.2%\uff0c\u8aaa\u660e\u9019\u985e world model \u4e0d\u53ea\u662f\u5167\u5bb9\u751f\u6210\u5de5\u5177\uff0c\u4ea6\u958b\u59cb\u6210\u70ba embodied intelligence \u7684\u8cc7\u6599\u5f15\u64ce\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u6587\u4e2d\u4ea6\u63d0\u5230\u5b8c\u6574\u6d41\u7a0b\u6db5\u84cb dataset curation\u3001unified annotation pipeline\u3001single-step training\u3001sequential training\uff0c\u4ee5\u53ca\u914d\u5408 FlashAR \u8207 vLLM Integration \u7684 inference \u8a2d\u8a08\u3002\u6574\u9ad4\u8a0a\u606f\u5f88\u6e05\u695a\uff1a\u4f5c\u8005\u60f3\u8b49\u660e foundation world models \u53ef\u4ee5\u540c\u6642\u626e\u6f14 embodied world models \u8207\u5408\u6210\u8cc7\u6599\u57fa\u5efa\uff0c\u8b93\u6a5f\u68b0\u4eba\u8a13\u7df4\u4e0d\u518d\u53ea\u4f9d\u8cf4\u6602\u8cb4\u800c\u7a00\u7f3a\u7684\u771f\u4eba\u793a\u7bc4\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/arxiv.org\/pdf\/2607.11643\" rel=\"noopener noreferrer\" target=\"_blank\"><strong>Paper<\/strong><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u540c\u4e00\u500b\u6a21\u578b\u517c\u9867\u5834\u666f\u751f\u6210\u3001\u5f71\u50cf\u7de8\u8f2f\u540c\u6a5f\u68b0\u4eba\u64cd\u4f5c\u8cc7\u6599\u5408\u6210\uff0c\u91cd\u9ede\u5514\u53ea\u4fc2\u756b\u9762\u975a\uff0c\u800c\u4fc2\u591a\u8996\u89d2\u4e00\u81f4\u540c\u53ef\u76f4\u63a5\u9935\u53bb\u8a13\u7df4\u7b56\u7565\u3002<\/p>\n","protected":false},"author":8,"featured_media":10223,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ai_generated_summary":"","footnotes":""},"categories":[31,30,24,129,119,157,76,127,141,184,186,199,206],"tags":[],"class_list":["post-10224","post","type-post","status-publish","format-standard","hentry","category-video","category-image","category-aisoftware","category-txt2img","category-119","category-157","category-76","category-127","category-141","category-robotic","category-186","category-dataset-","category-xiaomi"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/10224","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/users\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/comments?post=10224"}],"version-history":[{"count":0,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/10224\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/10223"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=10224"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=10224"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=10224"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}