{"id":222,"date":"2025-12-25T14:48:02","date_gmt":"2025-12-25T06:48:02","guid":{"rendered":"https:\/\/1ow.top\/index.php\/2025\/12\/25\/%e5%8e%9f%e5%8a%9b%e7%81%b5%e6%9c%ba%e6%8f%90%e5%87%bageovla%ef%bc%9a%e8%ae%a9%e6%9c%ba%e5%99%a8%e4%ba%ba%e7%9c%8b%e6%87%82%e4%b8%89%e7%bb%b4%e4%b8%96%e7%95%8c%ef%bc%8c%e6%89%93%e7%a0%b42d%e8%a7%86\/"},"modified":"2025-12-25T14:48:02","modified_gmt":"2025-12-25T06:48:02","slug":"%e5%8e%9f%e5%8a%9b%e7%81%b5%e6%9c%ba%e6%8f%90%e5%87%bageovla%ef%bc%9a%e8%ae%a9%e6%9c%ba%e5%99%a8%e4%ba%ba%e7%9c%8b%e6%87%82%e4%b8%89%e7%bb%b4%e4%b8%96%e7%95%8c%ef%bc%8c%e6%89%93%e7%a0%b42d%e8%a7%86","status":"publish","type":"post","link":"https:\/\/1ow.top\/index.php\/2025\/12\/25\/%e5%8e%9f%e5%8a%9b%e7%81%b5%e6%9c%ba%e6%8f%90%e5%87%bageovla%ef%bc%9a%e8%ae%a9%e6%9c%ba%e5%99%a8%e4%ba%ba%e7%9c%8b%e6%87%82%e4%b8%89%e7%bb%b4%e4%b8%96%e7%95%8c%ef%bc%8c%e6%89%93%e7%a0%b42d%e8%a7%86\/","title":{"rendered":"\u539f\u529b\u7075\u673a\u63d0\u51faGeoVLA\uff1a\u8ba9\u673a\u5668\u4eba\u770b\u61c2\u4e09\u7ef4\u4e16\u754c\uff0c\u6253\u78342D\u89c6\u89c9\u67b7\u9501"},"content":{"rendered":"
\"\u539f\u529b\u7075\u673a\u63d0\u51faGeoVLA\uff1a\u8ba9\u673a\u5668\u4eba\u770b\u61c2\u4e09\u7ef4\u4e16\u754c\uff0c\u6253\u78342D\u89c6\u89c9\u67b7\u9501\"\/<\/div><\/figure>\n
\"\u539f\u529b\u7075\u673a\u63d0\u51faGeoVLA\uff1a\u8ba9\u673a\u5668\u4eba\u770b\u61c2\u4e09\u7ef4\u4e16\u754c\uff0c\u6253\u78342D\u89c6\u89c9\u67b7\u9501\"\/<\/div><\/figure>\n
\"\u539f\u529b\u7075\u673a\u63d0\u51faGeoVLA\uff1a\u8ba9\u673a\u5668\u4eba\u770b\u61c2\u4e09\u7ef4\u4e16\u754c\uff0c\u6253\u78342D\u89c6\u89c9\u67b7\u9501\"\/<\/div><\/figure>\n
\"\u539f\u529b\u7075\u673a\u63d0\u51faGeoVLA\uff1a\u8ba9\u673a\u5668\u4eba\u770b\u61c2\u4e09\u7ef4\u4e16\u754c\uff0c\u6253\u78342D\u89c6\u89c9\u67b7\u9501\"\/<\/div><\/figure>\n

\u5728\u5177\u8eab\u667a\u80fd\u7684\u6d6a\u6f6e\u4e2d\uff0cVLA\u6a21\u578b\u88ab\u89c6\u4e3a\u901a\u5f80\u901a\u7528\u673a\u5668\u4eba\u7684\u5feb\u8f66\u9053\u3002<\/p>\n

\u7136\u800c\uff0c\u968f\u7740\u7814\u7a76\u6df1\u5165\u5230\u975e\u7ed3\u6784\u5316\u73af\u5883\uff0c\u73b0\u6709VLA\u6a21\u578b\u9762\u4e34\u7740\u4e00\u4e2a\u4e25\u91cd\u7684\u7ef4\u5ea6\u7f3a\u9677\uff1a\u7a7a\u95f4\u5931\u660e\u3002<\/p>\n

\u76ee\u524d\uff0c\u5927\u591a\u6570VLA\u6a21\u578b\uff08\u6bd4\u5982 OpenVLA\u3001RT-2\u3001Pi0\u3001Pi05\uff09\u5355\u7eaf\u4f9d\u8d562D RGB\u56fe\u50cf\u4f5c\u4e3a\u89c6\u89c9\u8f93\u5165\uff0c\u5bfc\u81f4\u6a21\u578b\u773c\u4e2d\u7684\u4e16\u754c\u201c\u7eb8\u7247\u5316\u201d\uff0c\u4e25\u91cd\u7f3a\u4e4f\u6df1\u5ea6\u4fe1\u606f\u548c\u51e0\u4f55\u5148\u9a8c\uff1b\u7531\u6b64\u5e26\u6765\u7684\u540e\u679c\u662f\uff1a<\/p>\n

Dexmal \u539f\u529b\u7075\u673a\u4f5c\u8005\u56e2\u961f\u63d0\u51fa\u4e00\u79cd\u5168\u65b0\u7684VLA\u6846\u67b6GeoVLA\uff0c\u5b83\u5728\u4fdd\u7559\u73b0\u6709\u89c6\u89c9-\u8bed\u8a00\u6a21\u578b\uff08VLM\uff09\u7684\u9884\u8bad\u7ec3\u80fd\u529b\u7684\u540c\u65f6\uff0c\u91c7\u7528\u4e86\u4e00\u79cd\u4f18\u96c5\u7684\u53cc\u6d41\u67b6\u6784\uff08Dual-path Architecture\uff09\u3002<\/p>\n

\u5177\u4f53\u800c\u8a00\uff0cGeoVLA\u5728\u4fdd\u7559VLM\u5f3a\u5927\u7684\u8bed\u4e49\u7406\u89e3\u80fd\u529b\u7684\u540c\u65f6\uff0c\u5f15\u5165\u4e13\u7528\u7684\u70b9\u4e91\u5d4c\u5165\u7f51\u7edc PEN \u548c\u7a7a\u95f4\u611f\u77e5\u52a8\u4f5c\u4e13\u5bb63DAE\uff0c\u76f4\u63a5\u5229\u7528\u6df1\u5ea6\u56fe\u751f\u6210\u7684\u70b9\u4e91\u6570\u636e\uff0c\u8d4b\u4e88\u673a\u5668\u4eba\u771f\u6b63\u7684\u4e09\u7ef4\u51e0\u4f55\u611f\u77e5\u80fd\u529b\u3002<\/p>\n

\u8fd9\u4e00\u8bbe\u8ba1\u4e0d\u4ec5\u5728\u4eff\u771f\u73af\u5883\u4e2d\u53d6\u5f97SOTA\uff0c\u66f4\u5728\u771f\u5b9e\u4e16\u754c\u7684\u9c81\u68d2\u6027\u6d4b\u8bd5\u4e2d\uff0c\u7279\u522b\u662f\u5728\u89c6\u89d2\u6539\u53d8\u548c\u7269\u4f53\u5c3a\u5ea6\u53d8\u5316\u7684\u6781\u7aef\u6761\u4ef6\u4e0b\uff0c\u5c55\u73b0\u51fa\u60ca\u4eba\u7684\u9002\u5e94\u529b\u3002<\/p>\n

\u5e38\u89c1\u7684\u505a\u6cd5\u8bd5\u56fe\u8ba9\u4e00\u4e2aVLM\u65e2\u61c2\u8bed\u4e49\u53c8\u61c2\u51e0\u4f55\uff0c\u8fd9\u5f80\u5f80\u987e\u6b64\u5931\u5f7c\uff1bGeoVLA\u7684\u6838\u5fc3\u903b\u8f91\u662f\u9009\u62e9\u628a\u4efb\u52a1\u89e3\u8026\uff1a<\/p>\n

\u8ba9 VLM \u8d1f\u8d23\u201c\u770b\u61c2\u662f\u4ec0\u4e48\u201d\uff0c\u8ba9\u70b9\u4e91\u7f51\u7edc\u8d1f\u8d23\u201c\u770b\u6e05\u5728\u54ea\u91cc\u201d\u3002<\/p>\n

GeoVLA \u662f\u4e00\u4e2a\u5168\u65b0\u7684\u7aef\u5230\u7aef\u6846\u67b6\uff0c\u5176\u6d41\u7a0b\u5305\u542b\u4e09\u4e2a\u5173\u952e\u7ec4\u4ef6\u7684\u534f\u540c\u5de5\u4f5c\uff1a<\/p>\n

\u539f\u59cb\u6df1\u5ea6\u56fe\u5f80\u5f80\u5305\u542b\u5927\u91cf\u566a\u58f0\uff0c\u4e14\u6570\u636e\u7a00\u758f\uff0c\u76f4\u63a5\u4f5c\u4e3a\u8f93\u5165\u6548\u679c\u4e0d\u4f73\u3002\u70b9\u4e91\u5d4c\u5165\u7f51\u7edc PEN \u4e13\u4e3a\u673a\u5668\u4eba\u64cd\u4f5c\u8bbe\u8ba1\uff0c\u91c7\u7528\u53cc\u8def\u5f84\u67b6\u6784\u6765\u63d0\u53d6\u5e72\u51c0\u4e14\u7d27\u51d1\u7684\u51e0\u4f55\u7279\u5f81\uff1a<\/p>\n

\u7a7a\u95f4\u951a\u70b9\uff08Spatial Anchor\uff09\u8bbe\u8ba1\u662f PEN \u7684\u4e00\u5927\u4eae\u70b9\u3002<\/p>\n

\u4f5c\u8005\u56e2\u961f\u5e76\u6ca1\u6709\u7b80\u5355\u5730\u5bf9\u6240\u6709\u70b9\u4e91\u7279\u5f81\u8fdb\u884c\u5e73\u5747\u6c60\u5316\uff0c\u800c\u662f\u9009\u62e9\u5bf9\u5e94\u4e8e\u672b\u7aef\u6267\u884c\u5668\u5750\u6807\u539f\u70b9\u7684 Token \u4f5c\u4e3a\u201c\u951a\u70b9\u201d\u3002<\/p>\n

\u8fd9\u79cd\u4ee5\u201c\u624b\u201d\u4e3a\u4e2d\u5fc3\u7684\u89c6\u89d2\u8bbe\u8ba1\uff0c\u8ba9\u6a21\u578b\u80fd\u591f\u663e\u5f0f\u5730\u5efa\u6a21\u201c\u624b\u201d\u4e0e\u201c\u7269\u4f53\u201d\u4e4b\u95f4\u7684\u51e0\u4f55\u5173\u7cfb\uff0c\u5927\u5e45\u63d0\u5347\u64cd\u4f5c\u7cbe\u5ea6\u3002<\/p>\n

\u7279\u5f81\u63d0\u53d6\u53ea\u662f\u7b2c\u4e00\u6b65\uff0c\u5982\u4f55\u6709\u6548\u878d\u5408RGB\u7684\u8bed\u4e49\u4fe1\u606f\u548c\u70b9\u4e91\u7684\u51e0\u4f55\u4fe1\u606f\uff0c\u5b9e\u73b01+1\uff1e2\u7684\u6548\u679c\uff0c\u662f\u591a\u6a21\u6001\u7814\u7a76\u5f53\u4e2d\u7684\u96be\u70b9\u3002<\/p>\n

\u4f5c\u8005\u56e2\u961f\u5728\u52a8\u4f5c\u751f\u6210\u7aef\u91c7\u7528\u57fa\u4e8e\u6269\u6563 Transformer \uff08DiTs\uff09\u7684\u67b6\u6784\uff0c\u5e76\u521b\u65b0\u6027\u5730\u5f15\u5165\u6df7\u5408\u4e13\u5bb6 \uff08MoE\uff09\u673a\u5236\u3002<\/p>\n

\u5b9e\u9a8c\u7ed3\u679c\u663e\u793a\uff0cGeoVLA\u5728\u4eff\u771f\u548c\u771f\u673a\u5b9e\u9a8c\u4e2d\u5747\u5c55\u73b0\u51fa\u5bf9\u4f20\u7edf2D VLA\u6a21\u578b\u7684\u538b\u5012\u6027\u4f18\u52bf\uff0c\u8bc1\u660e\u663e\u5f0f3D\u8868\u5f81\u5728\u590d\u6742\u64cd\u4f5c\u4e2d\u7684\u4e0d\u53ef\u66ff\u4ee3\u6027\u3002<\/p>\n

\u5728LIBERO\u57fa\u51c6\u6d4b\u8bd5\u4e2d\uff0cGeoVLA\u8d85\u8d8a\u6240\u6709\u4efb\u52a1\u5957\u4ef6\u3002<\/p>\n

\u5728\u6700\u5177\u6311\u6218\u6027\u7684 LIBERO-90\uff08\u957f\u7a0b\u591a\u4efb\u52a1\uff09\u4e2d\uff0cGeoVLA\u8fbe\u523097.7%\u7684\u6210\u529f\u7387\uff0c\u8d85\u8d8a\u4e4b\u524d\u7684SOTA\u65b9\u6cd5OpenVLA-OFT (95.3%) \u548c CogACT (93.2%)\u3002<\/p>\n

\u5728\u7269\u7406\u4eff\u771f\u66f4\u4e3a\u903c\u771f\u7684ManiSkill2\u4e2d\uff0cGeoVLA\u4f18\u52bf\u66f4\u52a0\u660e\u663e\uff0c\u5e73\u5747\u6210\u529f\u7387\u8fbe\u5230 77%\uff0c\u5927\u5e45\u9886\u5148Dita (66%) \u548cCogACT (69%)\u3002<\/p>\n

\u7279\u522b\u662f\u5728 PickClutterYCB \u8fd9\u79cd\u7269\u4f53\u5806\u53e0\u6742\u4e71\u3001\u906e\u6321\u4e25\u91cd\u7684\u4efb\u52a1\u4e2d\uff0cGeoVLA \u51ed\u501f\u70b9\u4e91\u5e26\u6765\u7684\u51e0\u4f55\u7406\u89e3\uff0c\u4fdd\u6301\u4e86\u6781\u9ad8\u7684\u64cd\u4f5c\u7cbe\u5ea6\u3002<\/p>\n

\u4f5c\u8005\u56e2\u961f\u4f7f\u7528 WidowX-250s \u673a\u68b0\u81c2\u8fdb\u884c\u4e86\u5e7f\u6cdb\u7684\u771f\u673a\u6d4b\u8bd5\uff1b\u5b9e\u9a8c\u88ab\u5206\u4e3a\u201c\u57fa\u7840\u4efb\u52a1\u201d\u548c\u201c3D \u611f\u77e5\u4efb\u52a1\u201d\u3002<\/p>\n

\u5728\u57df\u5185\u4efb\u52a1\u4e2d\uff0cGeoVLA \u5728\u57fa\u7840\u4efb\u52a1\u4e0a\u5e73\u5747\u6210\u529f\u7387 95.0%\uff0c\u5728 3D \u611f\u77e5\u4efb\u52a1\u4e0a\u4e3a 77.5%\uff0c\u603b\u4f53\u5e73\u5747 86.3%\uff0c\u5927\u5e45\u9886\u5148 Pi0 (57.5%) \u548c CogACT (76.3%)\u3002<\/p>\n

\u7279\u522b\u662f\u5728 Put Basketball \u548c Put Hairclip \u7b49\u9700\u8981\u7cbe\u786e\u7a7a\u95f4\u7406\u89e3\u7684\u4efb\u52a1\u4e2d\uff0cGeoVLA \u8868\u73b0\u51fa\u66f4\u597d\u7684\u9c81\u68d2\u6027\u3002<\/p>\n

\u66f4\u4ee4\u4eba\u5370\u8c61\u6df1\u523b\u7684\u662f GeoVLA \u5728\u5206\u5e03\u5916\uff08OOD\uff09\u573a\u666f\u4e0b\u7684\u9c81\u68d2\u6027\uff0c\u8fd9\u4e5f\u662f GeoVLA \u6700\u6838\u5fc3\u7684\u7a81\u7834\u70b9\uff1a<\/p>\n

\u25b3\uff08\u5de6\uff09\uff1a\u6295\u7bee\u4efb\u52a1\u53d8\u4f53\u7684\u8bc4\u6d4b\u7ed3\u679c\uff1b\uff08\u53f3\uff09\uff1a\u5957\u5a03\u4efb\u52a1\u53d8\u4f53\u7684\u8bc4\u6d4b\u7ed3\u679c<\/p>\n

\u53c2\u8003\u6587\u732e<\/p>\n

\u2022 Chengmeng Li, Junjie Wen, Yan Peng, Yaxin Peng, Feifei Feng, and Yichen Zhu. Pointvla: Injecting the 3d world into vision-language-action models. arXiv preprint arXiv:2503.07511, 2025.\u2022 Delin Qu, Haoming Song, Qizhi Chen, Yuanqi Yao, Xinyi Ye, Yan Ding, Zhigang Wang, JiaYuan Gu, Bin Zhao,Dong Wang, and Xuelong Li. Spatialvla: Exploring spatial representations for visual-language-action model. In Robotics: Science and Systems (RSS), 2025.\u2022 Zhi Hou, Tianyi Zhang, Yuwen Xiong, Haonan Duan, Hengjun Pu, Ronglei Tong, Chengyang Zhao, Xizhou Zhu,Yu Qiao, Jifeng Dai, and Yuntao Chen. Dita: Scaling diffusion transformer for generalist vision-language-action policy. arXiv preprint arXiv:2503.19757, 2025.\u2022 {Moo Jin} Kim, Karl Pertsch, Siddharth Karamcheti, Ted Xiao, Ashwin Balakrishna, Suraj Nair, Rafael Rafailov, Ethan Foster, Grace Lam, Pannag Sanketi, Quan Vuong, Thomas Kollar, Benjamin Burchfiel, Russ Tedrake, Dorsa Sadigh, Sergey Levine, Percy Liang, and Chelsea Finn. Openvla: An open-source vision-language-action model. arXiv preprint arXiv:2406.09246\u2022 Li Qixiu, Liang Yaobo, Wang Zeyu, Luo Lin, Chen Xi, Liao Mozheng, Wei Fangyun, Deng Yu, Xu Sicheng, Zhang Yizhong, and others. CogACT: A Foundational Vision-Language-Action Model for Synergizing Cognition and Action in Robotic Manipulation. arXiv preprint arXiv:2411.19650.<\/p>\n

\u8bba\u6587\u540d\u79f0\uff1aGeoVLA: Empowering 3D Representation in Vision-Language-Action Models\u8bba\u6587\u94fe\u63a5\uff1ahttps:\/\/arxiv.org\/html\/2508.09071v2<\/p>\n

\u9879\u76ee\u4e3b\u9875\uff1ahttps:\/\/linsun449.github.io\/GeoVLA\/<\/p>\n

\n

\u7248\u6743\u58f0\u660e\uff1a<\/strong><\/p>\n

\u672c\u535a\u5ba2\u90e8\u5206\u5185\u5bb9\u4e3a\u8f6c\u8f7d\u6587\u7ae0\uff0c\u65e8\u5728\u5206\u4eab\u6709\u4ef7\u503c\u7684\u4fe1\u606f\uff0c\u7248\u6743\u5f52\u539f\u4f5c\u8005\u6240\u6709\u3002<\/p>\n

\u8f6c\u8f7d\u4ec5\u4e3a\u4e2a\u4eba\u5b66\u4e60\u4e0e\u4ea4\u6d41\u76ee\u7684\uff0c\u4e0d\u5bf9\u6587\u7ae0\u89c2\u70b9\u8d1f\u8d23\uff0c\u4ea6\u4e0d\u7528\u4e8e\u4efb\u4f55\u5546\u4e1a\u7528\u9014\u3002<\/p>\n

\u5982\u6d89\u53ca\u7248\u6743\u95ee\u9898\uff0c\u8bf7\u8054\u7cfb\u672c\u4eba\u5220\u9664\u3002<\/p>\n<\/blockquote>\n","protected":false},"excerpt":{"rendered":"

\u5728\u5177\u8eab\u667a\u80fd\u7684\u6d6a\u6f6e\u4e2d\uff0cVLA\u6a21\u578b\u88ab\u89c6\u4e3a\u901a\u5f80\u901a\u7528\u673a\u5668\u4eba\u7684\u5feb\u8f66\u9053\u3002<\/p>\n

\u7136\u800c\uff0c\u968f\u7740\u7814\u7a76\u6df1\u5165\u5230\u975e\u7ed3\u6784\u5316\u73af\u5883\uff0c\u73b0\u6709VLA\u6a21\u578b\u9762\u4e34\u7740\u4e00\u4e2a\u4e25\u91cd\u7684\u7ef4\u5ea6\u7f3a\u9677\uff1a\u7a7a\u95f4\u5931\u660e\u3002<\/p>\n

\u76ee\u524d\uff0c\u5927\u591a\u6570VLA\u6a21\u578b\uff08\u6bd4\u5982 OpenVLA\u3001RT-2\u3001Pi0\u3001Pi05\uff09\u5355\u7eaf\u4f9d\u8d562D RGB\u56fe\u50cf\u4f5c\u4e3a\u89c6\u89c9\u8f93\u5165\uff0c\u5bfc\u81f4\u6a21\u578b\u773c\u4e2d\u7684\u4e16\u754c\u201c\u7eb8\u7247\u5316\u201d\uff0c\u4e25\u91cd\u7f3a\u4e4f\u6df1\u5ea6\u4fe1\u606f\u548c\u51e0\u4f55\u5148\u9a8c\uff1b\u7531\u6b64\u5e26\u6765\u7684\u540e\u679c\u662f\uff1a<\/p>\n

Dexmal \u539f\u529b\u7075\u673a\u4f5c\u8005\u56e2\u961f\u63d0\u51fa\u4e00\u79cd\u5168\u65b0\u7684VLA\u6846\u67b6G<\/p>\n","protected":false},"author":1,"featured_media":218,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[7,8,9],"class_list":["post-222","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-xwzx","tag-vla","tag-8","tag-9"],"_links":{"self":[{"href":"https:\/\/1ow.top\/index.php\/wp-json\/wp\/v2\/posts\/222","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/1ow.top\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/1ow.top\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/1ow.top\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/1ow.top\/index.php\/wp-json\/wp\/v2\/comments?post=222"}],"version-history":[{"count":0,"href":"https:\/\/1ow.top\/index.php\/wp-json\/wp\/v2\/posts\/222\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/1ow.top\/index.php\/wp-json\/wp\/v2\/media\/218"}],"wp:attachment":[{"href":"https:\/\/1ow.top\/index.php\/wp-json\/wp\/v2\/media?parent=222"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/1ow.top\/index.php\/wp-json\/wp\/v2\/categories?post=222"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/1ow.top\/index.php\/wp-json\/wp\/v2\/tags?post=222"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}