{"id":13605,"date":"2026-10-20T16:51:13","date_gmt":"2026-10-20T08:51:13","guid":{"rendered":"https:\/\/infernews.com\/blog\/scalenav\/"},"modified":"2026-10-20T16:52:25","modified_gmt":"2026-10-20T08:52:25","slug":"scalenav","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/scalenav\/","title":{"rendered":"ScaleNav \u628a\u5f71\u7247\u8b8a\u6210\u5c0e\u822a\u80fd\u529b"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/10\/ScaleNav.jpg\" alt=\"ScaleNav navigation: real robot on the left, simulation on the right\"\/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">\u7db2\u4e0a\u7b2c\u4e00\u8996\u89d2\u5f71\u7247 seldom \u9644\u5e36\u6a5f\u5668\u4eba\u52d5\u4f5c\uff0c\u4f46\u7576\u4e2d\u860a\u85cf\u5927\u91cf\u300c\u6cbf\u8d70\u5eca\u524d\u9032\u3001\u907f\u958b\u969c\u7919\u3001\u8f49\u5165\u76ee\u6a19\u623f\u9593\u300d\u7684\u8996\u89ba\u898f\u5f8b\u3002ScaleNav \u662f\u4e00\u500b\u7814\u7a76\u578b\u5c0e\u822a\u6a21\u578b\uff0c\u8a66\u5716\u628a\u9019\u4e9b action-free \u9810\u8a13\u7df4\u6240\u5f97\u5230\u7684 navigation priors\uff0c\u8f49\u79fb\u5230\u6709\u52d5\u4f5c\u6a19\u7c64\u7684\u6a21\u64ec\u8ecc\u8de1\uff0c\u518d\u7528\u65bc language-conditioned local navigation\uff1a\u6309\u6587\u5b57\u5c0b\u627e\u521d\u59cb\u756b\u9762\u53ef\u898b\u7684\u76ee\u6a19\u4e26\u505c\u4e0b\u4f86\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u6a21\u578b\u5c6c\u5171\u4eab\u5f0f Diffusion Transformer\uff0c\u6703\u6839\u64da\u6307\u4ee4\u3001\u76ee\u524d RGB \u5f71\u50cf\u53ca\u8a18\u61b6\uff0c\u5171\u540c\u9810\u6e2c\u672a\u4f86 video latents \u548c action chunks\u3002\u6d41\u7a0b\u5206\u4e09\u968e\u6bb5\uff1a\u5148\u5f9e\u4e92\u806f\u7db2\u7b2c\u4e00\u8996\u89d2\u5f71\u7247\u5b78\u7fd2\u76ee\u6a19\u5c0e\u5411\u8996\u89ba\u52d5\u614b\uff0c\u518d\u9032\u884c action alignment\uff0c\u6700\u5f8c joint fine-tuning\uff0c\u540c\u6642\u4fdd\u7559\u5f71\u7247\u56de\u653e\u4e26\u5fae\u8abf\u5171\u4eab\u9aa8\u5e72\u8207\u8f38\u51fa\u982d\u3002<\/p>\n\n\n<div\n    \tclass=\"align wp-block-vpb-video\"    id='vpbpVideoPlayer-1'\n    data-attributes='{&quot;source&quot;:&quot;https:\\\/\\\/zju3dv.github.io\\\/ScaleNav\\\/assets\\\/demo-6.mp4&quot;,&quot;repeat&quot;:true,&quot;autoplay&quot;:true,&quot;muted&quot;:true,&quot;align&quot;:&quot;&quot;,&quot;poster&quot;:&quot;&quot;,&quot;controls&quot;:{&quot;play-large&quot;:true,&quot;restart&quot;:false,&quot;rewind&quot;:true,&quot;play&quot;:true,&quot;fast-forward&quot;:true,&quot;progress&quot;:true,&quot;current-time&quot;:true,&quot;duration&quot;:false,&quot;mute&quot;:true,&quot;volume&quot;:true,&quot;pip&quot;:false,&quot;airplay&quot;:false,&quot;captions&quot;:false,&quot;settings&quot;:true,&quot;download&quot;:false,&quot;fullscreen&quot;:true},&quot;captions&quot;:[],&quot;ratio&quot;:&quot;16:9&quot;,&quot;accentColor&quot;:&quot;&quot;,&quot;lazyLoad&quot;:false,&quot;width&quot;:&quot;100%&quot;,&quot;radius&quot;:&quot;0px&quot;,&quot;resetOnEnd&quot;:false,&quot;autoHideControl&quot;:true,&quot;isSetup&quot;:false}'\n><\/div>\n\n\n<p class=\"wp-block-paragraph\">ScaleNav \u8207\u540c\u985e\u505a\u6cd5\u7684\u4e3b\u8981\u53d6\u6368\uff0c\u662f\u6bcb\u9808\u5148\u6536\u96c6\u5927\u91cf\u5b8c\u6574\u6a5f\u5668\u4eba\u793a\u7bc4\uff0c\u4fbf\u80fd\u501f\u52a9\u4e92\u806f\u7db2\u5f71\u7247\u5efa\u7acb\u5c0e\u822a\u5148\u9a57\uff1b\u4ee3\u50f9\u662f\u4ecd\u8981\u4f9d\u8cf4\u6a21\u64ec\u8ecc\u8de1\u3001\u52d5\u4f5c\u5c0d\u9f4a\u53ca\u591a\u968e\u6bb5\u8a13\u7df4\u3002\u5b98\u65b9\u6a21\u578b\u7d04 10.02 GB\uff0c\u63a8\u8ad6\u53ca\u8a55\u4f30\u5efa\u8b70\u4f7f\u7528\u652f\u63f4 BF16 \u7684 NVIDIA GPU\u300124 GB GPU \u8a18\u61b6\u9ad4\u8207 64 GB \u7cfb\u7d71\u8a18\u61b6\u9ad4\uff0c\u4e26\u6309\u6587\u4ef6\u914d\u7f6e Linux\u3001Habitat-GS \u548c\u5834\u666f\u8f49\u63db\u5de5\u5177\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u7d50\u679c\u6db5\u84cb HabitatGS-LN\u3001Short-Horizon OVON \u8207\u771f\u5be6\u56db\u8db3\u6a5f\u68b0\u4eba\uff1b\u516c\u958b\u8cc7\u6599\u5217\u51fa ScaleNav \u5728\u5169\u9805\u6a21\u64ec\u8a55\u4f30\u7684\u6210\u529f\u7387\u5206\u5225\u70ba 94.9% \u53ca 86.5%\uff0c\u771f\u5be6\u6a5f\u68b0\u4eba\u5247\u5728 60 \u6b21\u8a66\u9a57\u4e2d\u9304\u5f97 83.3% \u6210\u529f\u7387\u3002\u5718\u968a\u4ea6\u6574\u7406 gallery \u5c0e\u822a\u5f71\u7247\u3001\u8ecc\u8de1\u3001\u76ee\u6a19\u8207\u7d50\u679c\uff0c\u4ee5\u53ca summary.json \u5167\u7684 SR\u3001SPL \u548c\u9010\u4efb\u52d9\u6210\u7e3e\uff0c\u65b9\u4fbf\u6bd4\u8f03\u8207\u91cd\u73fe\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5f71\u7247\u9810\u8a13\u7df4\u4e0d\u5fc5\u9644\u5e36\u9010\u683c\u52d5\u4f5c\u6a19\u7c64<\/li>\n\n\n\n<li>\u540c\u4e00\u6a21\u578b\u8655\u7406\u5f71\u7247\u52d5\u614b\u8207\u52d5\u4f5c\u751f\u6210<\/li>\n\n\n\n<li>\u652f\u63f4\u6a21\u64ec\u74b0\u5883\u53ca\u771f\u5be6\u56db\u8db3\u6a5f\u68b0\u4eba<\/li>\n\n\n\n<li>\u63d0\u4f9b\u6a21\u578b\u3001\u8cc7\u6599\u3001\u8a55\u4f30\u7d50\u679c\u8207\u64cd\u4f5c\u6587\u4ef6<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">\u9019\u5957\u65b9\u6cd5\u8f03\u9069\u5408\u7814\u7a76 embodied navigation\u3001 robotics \u8207 VLA\uff08Vision-Language-Action\uff09\u7684\u5718\u968a\uff1b\u8a55\u4f30\u6578\u5b57\u96d6\u4eae\u773c\uff0c\u4ecd\u9808\u7559\u610f\u5404\u65b9\u6cd5\u63a1\u7528\u4e0d\u540c\u8a2d\u5b9a\uff0c\u8de8\u6a21\u578b\u6bd4\u8f03\u61c9\u540c\u6642\u6838\u5c0d episode \u6578\u3001\u5834\u666f\u53ca fine-tuning \u689d\u4ef6\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/zju3dv.github.io\/ScaleNav\/\" rel=\"noopener noreferrer\"><strong>\u9805\u76ee\u4e3b\u9801<\/strong><\/a> \u00b7 <a href=\"https:\/\/github.com\/MachEmbodied\/ScaleNav\" rel=\"noopener noreferrer\"><strong>GitHub<\/strong><\/a> \u00b7 <a href=\"https:\/\/huggingface.co\/Yangyihui\/ScaleNav\" rel=\"noopener noreferrer\"><strong>\u6a21\u578b<\/strong><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>ScaleNav \u5148\u5f9e\u4e92\u806f\u7db2\u7b2c\u4e00\u8996\u89d2\u5f71\u7247\u5b78\u7fd2\u8def\u5f91\u898f\u5f8b\uff0c\u518d\u8f49\u6210\u6a5f\u5668\u4eba\u52d5\u4f5c\uff0c\u9762\u5c0d\u8a9e\u8a00\u6307\u5b9a\u76ee\u6a19\u7684\u5c40\u90e8\u5c0e\u822a\u3002<\/p>\n","protected":false},"author":8,"featured_media":13604,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"wpai_generated_summary":"","wpai_meta_description":"","footnotes":""},"categories":[133,162,76,149,119,127,179,31,30,204,184,38,168],"tags":[],"class_list":["post-13605","post","type-post","status-publish","format-standard","hentry","category-133","category-ai-productions","category-76","category-149","category-119","category-127","category-nvidia","category-video","category-image","category-visionlanguageaction","category-robotic","category-38","category-linux"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/13605","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=13605"}],"version-history":[{"count":1,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/13605\/revisions"}],"predecessor-version":[{"id":13607,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/13605\/revisions\/13607"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/13604"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=13605"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=13605"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=13605"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}