{"id":10591,"date":"2026-08-01T00:37:39","date_gmt":"2026-07-31T16:37:39","guid":{"rendered":"https:\/\/infernews.com\/blog\/humanclaw\/"},"modified":"2026-08-01T00:37:39","modified_gmt":"2026-07-31T16:37:39","slug":"humanclaw","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/humanclaw\/","title":{"rendered":"HumanCLAW \u76f4\u6307 VLM \u8eab\u9ad4\u611f\u7f3a\u53e3"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/08\/pasted-b65b7ab06db6.jpg\" alt=\"HumanCLAW teaser\"><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">\u756b\u9762\u7747\u5f97\u660e\uff0c\u4e0d\u7b49\u65bc\u8eab\u9ad4\u8b58\u5f97\u90c1\u5f97\u5571\u3002HumanCLAW \u628a Vision-Language Models\uff08VLMs\uff09\u653e\u9032\u4e00\u500b\u9589\u74b0\u4eba\u5f62\u884c\u52d5\u6e2c\u8a66\u74b0\u5883\uff0c\u96c6\u4e2d\u91cf\u5ea6\u6a21\u578b\u6bcf\u500b\u77ac\u9593\u61c9\u8a72\u505a\u54ea\u500b\u52d5\u4f5c\uff0c\u800c\u4e0d\u662f\u628a\u5931\u6557\u5168\u6578\u6b78\u548e\u65bc\u4f4e\u5c64\u99ac\u9054\u63a7\u5236\u3002\u5b83\u5c6c\u65bc<strong>\u8a55\u6e2c\u6846\u67b6\u517c\u57fa\u6e96\u6e2c\u8a66\u9805\u76ee<\/strong>\uff0c\u8655\u7406\u7684\u662f VLM \u5728\u5177\u8eab\u5834\u666f\u4e2d\u7684\u884c\u52d5\u6c7a\u7b56\u80fd\u529b\uff0c\u5230\u5e95\u6709\u6c92\u6709\u8db3\u5920\u300c\u8eab\u9ad4\u611f\u300d\u53bb\u5b8c\u6210\u627e\u8def\u3001\u79fb\u52d5\u8207\u4e92\u52d5\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u5462\u500b\u8a2d\u8a08\u6700\u503c\u5f97\u7559\u610f\u7684\u5730\u65b9\uff0c\u662f\u5b83\u628a action decision-making \u8207 low-level motor execution \u5206\u958b\u3002\u6bcf 0.5 \u79d2\uff0c\u51cd\u7d50\u7684 VLM \u53ea\u9700\u8981\u6839\u64da\u7b2c\u4e00\u8eab\u8996\u89d2\u3001\u6307\u4ee4\u3001\u6280\u80fd\u5217\u8868\u8207\u6b77\u53f2\u5167\u5bb9\uff0c\u63d0\u51fa\u4e00\u500b atomic whole-body skill\uff1b\u5f8c\u9762\u7684 verifier\u3001motion generator \u540c half-physics simulator \u518d\u8ca0\u8cac\u9a57\u8b49\u3001\u5b89\u5168\u904e\u6ffe\u8207\u9023\u7e8c\u52d5\u4f5c\u57f7\u884c\uff0c\u4ee4\u63a5\u89f8\u3001\u78b0\u649e\u3001\u91cd\u529b\u7b49\u7269\u7406\u5f8c\u679c\u4ecd\u7136\u4fdd\u7559\u4e0b\u4f86\uff0c\u4f46\u5e73\u8861\u5931\u8aa4\u8207\u52d5\u4f5c\u8ffd\u8e64\u8aa4\u5dee\u6703\u88ab\u76e1\u91cf\u6392\u9664\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">HumanCLAW-Bench \u5247\u5728\u9019\u500b\u6846\u67b6\u4e4b\u4e0a\u63d0\u4f9b 1,218 \u500b\u9577\u6642\u7a0b find\u2013navigate\u2013interact episodes\uff0c\u8986\u84cb 41 \u500b\u5ba4\u5167\u5834\u666f\u3002\u6578\u5b57\u76f8\u7576\u76f4\u63a5\uff1a\u4e5d\u500b\u6700\u5148\u9032 VLM \u5168\u90e8\u672a\u80fd\u89e3\u6c7a\u9019\u5957\u57fa\u6e96\uff0c\u6700\u4f73\u6210\u7e3e\u53ea\u6709 16.8% success rate\uff0c\u53cd\u6620\u554f\u984c\u4e0d\u5728\u55ae\u6b21\u8fa8\u8b58\uff0c\u800c\u5728\u6a21\u578b\u6301\u7e8c\u8ffd\u8e64\u81ea\u8eab\u4f4d\u7f6e\u3001\u5224\u65b7\u662f\u5426\u5230\u9054\u76ee\u6a19\uff0c\u4ee5\u53ca\u7406\u89e3\u81ea\u5df1\u6709\u6c92\u6709\u649e\u4e0a\u74b0\u5883\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u628a\u9ad8\u5c64\u6c7a\u7b56\u540c\u4f4e\u5c64\u52d5\u4f5c\u5206\u96e2\uff0c\u8f03\u6613\u7747\u6e05 VLM \u771f\u6b63\u5f31\u9ede<\/li>\n<li>\u4fdd\u7559\u771f\u5be6\u7269\u7406\u5f8c\u679c\uff0c\u5514\u6703\u56e0\u70ba\u7d14\u7b26\u865f\u5316\u74b0\u5883\u800c\u9ad8\u4f30\u80fd\u529b<\/li>\n<li>HumanCLAW-Bench \u8457\u91cd\u9577\u6642\u7a0b\u3001\u7b2c\u4e00\u8eab\u8996\u89d2\u3001\u9023\u7e8c\u4e92\u52d5\u4efb\u52d9<\/li>\n<li>\u76ee\u524d\u516c\u958b\u8cc7\u8a0a\u986f\u793a\u7a0b\u5f0f\u78bc\u8207 benchmark \u4ecd\u5728\u6e96\u5099\u91cb\u51fa<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">\u5c0d\u7814\u7a76 embodied AI\u3001Computer-use agents \u5ef6\u4f38\u65b9\u5411\u3001VLM \u8a55\u6e2c\u65b9\u6cd5\u7684\u4eba\u4f86\u8aaa\uff0c\u5462\u500b\u9805\u76ee\u6709\u53c3\u8003\u50f9\u503c\uff0c\u5c24\u5176\u9069\u5408\u7528\u4f86\u6aa2\u67e5\u6a21\u578b\u662f\u5426\u5177\u5099 closed-loop spatial action intelligence\uff0c\u800c\u4e0d\u53ea\u662f\u8b58\u63cf\u8ff0\u756b\u9762\u3002\u73fe\u968e\u6bb5\u8f03\u5927\u7684\u9650\u5236\u4e5f\u5f88\u6e05\u695a\uff1aGitHub \u5132\u5b58\u5eab\u5c1a\u672a\u6b63\u5f0f\u653e\u51fa harness\u3001motion generator weights\u3001half-physics simulation environment \u8207\u5b8c\u6574\u8a55\u6e2c\u5167\u5bb9\uff0c\u66ab\u6642\u4e3b\u8981\u4ecd\u662f\u900f\u904e project page\u3001paper \u540c leaderboard \u7406\u89e3\u65b9\u6cd5\u8207\u7d50\u679c\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/human-claw.github.io\/\" rel=\"noopener noreferrer\"><strong>\u9805\u76ee\u4e3b\u9801<\/strong><\/a> \u00b7 <a href=\"https:\/\/github.com\/Human-CLAW\/HumanCLAW\" rel=\"noopener noreferrer\"><strong>GitHub<\/strong><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u7576 Vision-Language Models \u771f\u6b63\u8981\u63a7\u5236\u4e00\u500b\u6703\u78b0\u649e\u3001\u6703\u53d7\u91cd\u529b\u5f71\u97ff\u7684\u4eba\u5f62\u8eab\u9ad4\uff0c\u8868\u73fe\u9060\u6bd4\u756b\u9762\u7406\u89e3\u56f0\u96e3\u3002HumanCLAW \u628a\u554f\u984c\u62c6\u958b\u4f86\u91cf\u5ea6\uff0c\u76f4\u63a5\u63ed\u958b VLM \u884c\u52d5\u5224\u65b7\u7684\u5f31\u9ede\u3002<\/p>\n","protected":false},"author":8,"featured_media":10590,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ai_generated_summary":"","footnotes":""},"categories":[133,116,119,149,188,199,201],"tags":[],"class_list":["post-10591","post","type-post","status-publish","format-standard","hentry","category-133","category-agentic","category-119","category-149","category-meta","category-dataset-","category-skill-"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/10591","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=10591"}],"version-history":[{"count":0,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/10591\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/10590"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=10591"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=10591"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=10591"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}