{"id":6366,"date":"2025-10-16T12:39:49","date_gmt":"2025-10-16T04:39:49","guid":{"rendered":"https:\/\/nullthought.net\/?p=6366"},"modified":"2025-10-16T12:39:50","modified_gmt":"2025-10-16T04:39:50","slug":"%e5%8c%85%e5%90%ab%e6%95%b0%e6%8d%ae-%e6%a8%a1%e5%9e%8b-%e5%b7%a5%e7%a8%8b%e4%b8%89%e4%bd%8d%e4%b8%80%e4%bd%93%e5%86%85%e5%ae%b9%e7%9a%84%e6%9c%ba%e5%99%a8%e4%ba%ba%e5%ad%a6%e4%b9%a0%ef%bc%88robot-lea","status":"publish","type":"post","link":"https:\/\/nullthought.net\/?p=6366","title":{"rendered":"\u5305\u542b\u6570\u636e\/\u6a21\u578b\/\u5de5\u7a0b\u4e09\u4f4d\u4e00\u4f53\u5185\u5bb9\u7684\u673a\u5668\u4eba\u5b66\u4e60\uff08Robot Learning\uff09\u6559\u7a0b"},"content":{"rendered":"\n<p><strong><a href=\"https:\/\/huggingface.co\/papers\/2510.12403\" target=\"_blank\" rel=\"noreferrer noopener\">Robot Learning: A Tutorial<\/a><\/strong>\u662f\u4e00\u7bc7\u9762\u5411\u7814\u7a76\u8005\u4e0e\u5b9e\u8df5\u8005\u7684\u201c\u673a\u5668\u4eba\u5b66\u4e60\u201d\u6559\u7a0b\u578b\u7efc\u8ff0\uff0c\u4e3b\u5f20\u5728\u4e0d\u629b\u5f03\u4f20\u7edf\u52a8\u529b\u5b66\/\u63a7\u5236\u5b66\u6210\u679c\u7684\u524d\u63d0\u4e0b\uff0c\u4ee5\u6570\u636e\u9a71\u52a8\u7684\u5b66\u4e60\u8303\u5f0f\uff08RL\/BC \u4e0e\u901a\u7528\u8bed\u8a00\u6761\u4ef6\u5316\u7b56\u7565\uff09\u91cd\u5851\u4ece\u201c\u611f\u77e5\u5230\u52a8\u4f5c\u201d\u7684\u7aef\u5230\u7aef\u63a7\u5236\u6808\u3002\u4f5c\u8005\u4e0d\u4ec5\u4f53\u7cfb\u5316\u68b3\u7406\u4e86\u4f20\u7edf\u4e0e\u5b66\u4e60\u8303\u5f0f\u7684\u5206\u91ce\u3001\u4e92\u8865\u8def\u5f84\u4e0e\u73b0\u5b9e\u63a3\u8098\uff0c\u8fd8\u7ed9\u51fa\u4e86\u5927\u91cf\u53ef\u590d\u73b0\u7684 <strong><a href=\"https:\/\/huggingface.co\/lerobot\" target=\"_blank\" rel=\"noreferrer noopener\">lerobot <\/a><\/strong>\u4ee3\u7801\u793a\u4f8b\u4e0e\u6570\u636e\u683c\u5f0f\u89c4\u8303\uff08LeRobotDataset\uff09\uff0c\u4ee5\u964d\u4f4e\u65b0\u624b\u5165\u95e8\u4e0e\u5de5\u7a0b\u843d\u5730\u95e8\u69db\u3002\u5168\u7bc7\u7ed3\u6784\u4f9d\u6b21\u4e3a\uff1a\u5f15\u8a00\u4e0e\u6570\u636e\u96c6\u683c\u5f0f\uff1b\u7ecf\u5178\u673a\u5668\u4eba\u5b66\u4e0e\u5176\u5c40\u9650\uff1b\u5f3a\u5316\u5b66\u4e60\u5728\u673a\u5668\u4eba\u4e0a\u7684\u673a\u9047\u4e0e\u96be\u9898\uff1b\u6a21\u4eff\u5b66\u4e60\u4e0e\u751f\u6210\u6a21\u578b\uff08VAE\/\u6269\u6563\/Flow Matching\u3001ACT\u3001Diffusion Policy\uff09\uff1b\u63a8\u7406\u4f18\u5316\uff1b\u901a\u7528\u578b\u673a\u5668\u4eba\u7b56\u7565\uff08VLA\u3001\u03c00\u3001SmolVLA\uff09\uff1b\u7ed3\u8bba\u4e0e\u5c55\u671b\u3002<\/p>\n\n\n\n<p>\u6587\u7ae0\u4f5c\u8005\u4e3aFrancesco Capuano, Caroline Pascal, Adil Zouitine, Thomas Wolf, Michel Aractingi\uff0c\u6765\u81eaUniversity of Oxford, Hugging Face\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"362\" src=\"https:\/\/nullthought.net\/wp-content\/uploads\/2025\/10\/image-9-1024x362.png\" alt=\"\" class=\"wp-image-6367\" srcset=\"https:\/\/nullthought.net\/wp-content\/uploads\/2025\/10\/image-9-1024x362.png 1024w, https:\/\/nullthought.net\/wp-content\/uploads\/2025\/10\/image-9-300x106.png 300w, https:\/\/nullthought.net\/wp-content\/uploads\/2025\/10\/image-9-768x272.png 768w, https:\/\/nullthought.net\/wp-content\/uploads\/2025\/10\/image-9.png 1346w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\"><strong><a href=\"https:\/\/huggingface.co\/papers\/2510.12403\" target=\"_blank\" rel=\"noreferrer noopener\">Robot Learning: A Tutorial<\/a><\/strong><\/figcaption><\/figure>\n\n\n\n<p>\u4e00\u3001LeRobotDataset\uff1a\u4e3a\u673a\u5668\u4eba\u5b66\u4e60\u800c\u751f\u7684\u6570\u636e\u57fa\u5efa<br>\u4f5c\u8005\u9996\u5148\u5f3a\u8c03\uff1a\u968f\u7740\u591a\u6a21\u6001\u673a\u5668\u4eba\u6570\u636e\u7684\u5927\u91cf\u6d8c\u73b0\uff0c\u6807\u51c6\u5316\u3001\u9ad8\u541e\u5410\u3001\u53ef\u6269\u5c55\u7684\u6570\u636e\u7ec4\u7ec7\u662f\u63a8\u52a8\u201c\u4ece\u5355\u4efb\u52a1\u5230\u901a\u624d\u201d\u8dc3\u8fc1\u7684\u5173\u952e\u3002LeRobotDataset \u7684\u8bbe\u8ba1\u8981\u70b9\u5305\u62ec\uff1a<br>1\uff09\u4e09\u5927\u652f\u67f1\u5f0f\u5b58\u50a8\uff1a\u8868\u683c\u6570\u636e\uff08\u5173\u8282\/\u52a8\u4f5c\u7b49\u4f4e\u7ef4\u9ad8\u9891\uff0c\u5185\u5b58\u6620\u5c04+datasets\u5e93\uff09\u3001\u89c6\u89c9\u6570\u636e\uff08\u540c\u4e00 episode \u5e27\u62fc\u63a5\u4e3a MP4\uff0c\u6309\u76f8\u673a\u89c6\u89d2\u5206\u7ec4+\u5206\u76ee\u5f55\u51cf\u8f7b\u6587\u4ef6\u7cfb\u7edf\u538b\u529b\uff09\u3001\u5143\u6570\u636e\uff08JSON \u8bb0\u5f55\u7279\u5f81\u6a21\u5f0f\u3001\u5e27\u7387\u3001\u5f52\u4e00\u5316\u7edf\u8ba1\u4e0e episode \u8fb9\u754c\uff0c\u5145\u5f53\u201c\u5173\u7cfb\u5c42\u201d\u6765\u91cd\u5efa\u8de8\u6587\u4ef6\u7684\u7d22\u5f15\uff09\u3002<br>2\uff09\u628a\u771f\u5b9e\u5b58\u50a8\u4e0e\u7528\u6237 API \u89e3\u8026\uff1a\u5e95\u5c42\u96c6\u4e2d\u62fc\u63a5\u3001\u4e0a\u5c42\u4ee5\u7a97\u53e3\u5316\uff08delta_timestamps\uff09\u76f4\u63a5\u5582\u7ed9 PyTorch DataLoader\uff0c\u65e2\u652f\u6301\u79bb\u7ebf\u6279\u5904\u7406\u4e5f\u652f\u6301 Hub \u6d41\u5f0f\u8bfb\u53d6\uff0c\u63d0\u5347\u6253\u4e71\u5ea6\u4e0e\u541e\u5410\uff0880\u2013100 it\/s \u7ea7\u522b\uff09\uff0c\u8d34\u5408 BC\/RL \u8bad\u7ec3\u5bf9\u65f6\u5e8f\u6808\u5e27\u548c\u52a8\u4f5c\u5757\u7684\u9700\u6c42\u3002<br>3\uff09\u9762\u5411\u591a\u5f62\u6001\u4e0e\u53ef\u6269\u5c55\uff1a\u5df2\u8986\u76d6 SO-100\/ALOHA-2\/\u4eff\u771f\/\u4eba\u5f62\/\u81ea\u52a8\u9a7e\u9a76\u7b49\u591a\u673a\u4f53\u6570\u636e\uff0c\u4fc3\u8fdb\u53ef\u590d\u73b0\u5b9e\u9a8c\u4e0e\u793e\u533a\u5171\u4eab\u3002<\/p>\n\n\n\n<p>\u4e8c\u3001\u7ecf\u5178\u673a\u5668\u4eba\u5b66\u7684\u80fd\u529b\u4e0e\u8fb9\u754c<br>\u4f5c\u8005\u4ee5\u201c\u663e\u5f0f\u6a21\u578b vs \u9690\u5f0f\u6a21\u578b\u201d\u5207\u5165\uff1a\u663e\u5f0f\u6a21\u578b\u4f9d\u8d56\u7cbe\u51c6\u7684\u51e0\u4f55\/\u52a8\u529b\u5b66\/\u63a5\u89e6\u5efa\u6a21\u4e0e\u89c4\u5212\u2014\u8ddf\u8e2a\u2014\u63a7\u5236\u6d41\u6c34\u7ebf\uff1b\u9690\u5f0f\u6a21\u578b\u5c06\u8fd0\u52a8\u89c6\u4e3a\u7edf\u8ba1\u6620\u5c04\uff0c\u7531\u6570\u636e\u5b66\u4e60\u611f\u77e5\u5230\u52a8\u4f5c\u7684\u76f4\u63a5\u51fd\u6570\u3002\u6559\u7a0b\u901a\u8fc7\u201c\u5e73\u9762 2 \u81ea\u7531\u5ea6\u64cd\u4f5c\u81c2\u201d\u7684\u73a9\u5177\u4f8b\u5b50\uff0c\u8bf4\u660e\u6b63\/\u9006\u8fd0\u52a8\u5b66\uff08FK\/IK\uff09\u3001\u5fae\u5206\u9006\u8fd0\u52a8\u5b66\uff08diff-IK\uff09\u548c\u53cd\u9988\uff08P\/PI\/PID\u3001LQR\u3001MPC\uff09\u5728\u9759\u6001\u53ef\u63a7\u73af\u5883\u4e2d\u7684\u6709\u6548\u6027\uff0c\u4ee5\u53ca\u5728\u969c\u788d\u3001\u63a5\u89e6\u3001\u975e\u7ebf\u6027\/\u4e0d\u786e\u5b9a\u6270\u52a8\u4e0b\u7684\u8106\u5f31\u6027\u548c\u8c03\u53c2\u6210\u672c\u3002\u66f4\u5173\u952e\u7684\u662f\uff0c\u4f20\u7edf\u6d41\u6c34\u7ebf\u5728\u6a21\u5757\u62fc\u88c5\u3001\u8bef\u5dee\u7ea7\u8054\u3001\u4f20\u611f\u591a\u6a21\u6001\u878d\u5408\u4e0e\u8de8\u4efb\u52a1\u8fc1\u79fb\u4e0a\u6210\u672c\u9ad8\u3001\u590d\u7528\u6027\u5dee\uff1b\u540c\u65f6\uff0c\u6469\u64e6\/\u987a\u5e94\/\u53ef\u53d8\u5f62\u4f53\u7b49\u73b0\u8c61\u7684\u7b80\u5316\u5efa\u6a21\u9650\u5236\u4e86\u771f\u5b9e\u4e16\u754c\u6027\u80fd\u3002<\/p>\n\n\n\n<p>\u4e09\u3001\u4e3a\u4f55\u8f6c\u5411\u5b66\u4e60\u5f0f\u673a\u5668\u4eba\uff08\u4ee5 RL \u4e3a\u4f8b\uff09<br>\u5b66\u4e60\u8303\u5f0f\u7684\u4f18\u52bf\u5728\u4e8e\uff1a<br>1\uff09\u5355\u4f53\u5316\u7684\u201c\u611f\u77e5\u2192\u52a8\u4f5c\u201d\u7b56\u7565\uff0c\u51cf\u5c11\u8106\u5f31\u63a5\u53e3\uff1b<br>2\uff09\u5929\u7136\u517c\u5bb9\u9ad8\u7ef4\u591a\u6a21\u6001\u8f93\u5165\uff08\u89c6\u89c9\/\u89e6\u89c9\/\u97f3\u9891\/\u672c\u4f53\u611f\u53d7\u7b49\uff09\uff1b<br>3\uff09\u4e0d\u4f9d\u8d56\u663e\u5f0f\u52a8\u529b\u5b66\u6a21\u578b\uff0c\u53ef\u76f4\u63a5\u7528\u4ea4\u4e92\u6570\u636e\u8fed\u4ee3\uff1b<br>4\uff09\u968f\u6570\u636e\u89c4\u6a21\u63d0\u5347\u800c\u53ef\u6269\u5c55\u3002\u6559\u7a0b\u7528\u5230\u6807\u51c6\u7684 MDP \u6846\u67b6\u4e0e\u56de\u62a5\u6700\u5927\u5316\u76ee\u6807\uff0c\u6982\u8ff0\u4e86\u4ef7\u503c\u51fd\u6570\/\u7b56\u7565\u4f18\u5316\u5173\u7cfb\u4e0e\u4e3b\u6d41\u7b97\u6cd5\uff08TRPO\/PPO\/SAC \u7b49\uff09\uff0c\u5e76\u4ee5\u5230\u8fbe\u2014\u653e\u7f6e\uff08\u64cd\u4f5c\uff09\u4e0e\u4fa7\u5411\u79fb\u52a8\uff08\u6b65\u6001\uff09\u793a\u4f8b\u5316\u201c\u5e8f\u8d2f\u51b3\u7b56\u201d\u7684\u672c\u8d28\u3002<\/p>\n\n\n\n<p>\u56db\u3001\u73b0\u5b9e\u673a\u5668\u4eba RL \u7684\u4e24\u5927\u75db\u70b9\uff1a\u5b89\u5168\u4e0e\u6837\u672c\u6548\u7387<br>1\uff09\u5b89\u5168\u4e0e\u4eba\u529b\uff1a\u65e9\u671f\u7b56\u7565\u63a2\u7d22\u5f80\u5f80\u201c\u778e\u201d\uff0c\u53ef\u80fd\u89e6\u53d1\u81ea\u649e\/\u8d85\u901f\/\u8d85\u529b\u77e9\u7b49\u98ce\u9669\uff0c\u540c\u65f6\u9700\u8981\u9891\u7e41\u4eba\u5de5\u590d\u4f4d\uff0c\u8bad\u7ec3\u8282\u594f\u6162\u3002<br>2\uff09\u6837\u672c\u6548\u7387\uff1a\u5373\u4fbf\u662f\u5f3a\u7b97\u6cd5\uff08\u5982 SAC\uff09\u4e5f\u5e38\u9700\u5927\u91cf\u4ea4\u4e92\u6b65\u6570\uff0c\u771f\u5b9e\u673a\u5668\u4eba\u4e0a\u4ee3\u4ef7\u9ad8\u6602\u3002<br>\u5e38\u89c1\u7f13\u89e3\u662f\u5728\u4eff\u771f\u4e2d\u8bad\u7ec3+\u57df\u968f\u673a\u5316\uff08DR\uff09\u8f6c\u5b9e\uff1a\u968f\u673a\u6469\u64e6\/\u8d28\u5fc3\/\u5149\u7167\u7b49\u53c2\u6570\u63d0\u9ad8\u9c81\u68d2\u6027\u3002\u7136\u800c DR \u9700\u8981\u624b\u5de5\u9009\u53c2\u4e0e\u5206\u5e03\u8bbe\u8ba1\uff0c\u71b5\u592a\u5c0f\u96be\u8f6c\u79fb\u3001\u592a\u5927\u5219\u8fc7\u6b63\u5219\uff1b\u8fd1\u671f\u65b9\u6cd5\u5c1d\u8bd5\u81ea\u52a8\u8c03 DR \u5206\u5e03\uff0c\u5982 AutoDR\uff08\u968f\u6027\u80fd\u62d3\u5bbd U(a,b) \u7684\u8fb9\u754c\uff09\u4e0e DORAEMON\uff08\u5b66\u5f97\u7684 Beta \u5206\u5e03\u3001\u5916\u5c42\u6700\u5927\u71b5+\u5185\u5c42\u6027\u80fd\u7ea6\u675f\uff09\uff0c\u53e6\u6709\u201c\u4ee5\u771f\u4fc3\u4eff\u201d\u7684\u5728\u7ebf\/\u79bb\u7ebf\u8f68\u8ff9\u914d\u51c6\u3002\u5c3d\u7ba1\u5982\u6b64\uff0c\u5f88\u591a\u63a5\u89e6\/\u53ef\u53d8\u5f62\u4efb\u52a1\u4eff\u771f\u4ecd\u96be\u9ad8\u4fdd\u771f\u3001\u7b97\u529b\u4ee3\u4ef7\u5927\uff1b\u66f4\u5e95\u5c42\u7684\u9650\u5236\u662f\u590d\u6742\u4efb\u52a1\u901a\u5e38\u96be\u4ee5\u7ed9\u51fa\u5bc6\u96c6\u5956\u52b1\uff0c\u7a00\u758f\u56de\u62a5\u663e\u8457\u653e\u6162\u5b66\u4e60\u3002\u4f5c\u8005\u56e0\u800c\u63d0\u5021\uff1a\u5c3d\u91cf\u951a\u5b9a\u5df2\u91c7\u96c6\u6f14\u793a\/\u7ecf\u9a8c\uff0c\u91c7\u7528\u6837\u672c\u9ad8\u6548\u7684\u79bb\u7ebf\/\u79bb\u7b56\u7565\u65b9\u6cd5\uff0c\u5e76\u7ed3\u5408\u201c\u4eba\u7c7b\u5728\u73af\u201d\u5e72\u9884\uff0c\u5df2\u5728 1\u20132 \u5c0f\u65f6\u5185\u628a\u771f\u5b9e\u4e16\u754c\u590d\u6742\u6293\u53d6\u64cd\u4f5c\u505a\u5230\u63a5\u8fd1\u6ee1\u5206\u6210\u529f\u7387\u3002<\/p>\n\n\n\n<p>\u4e94\u3001\u6a21\u4eff\u5b66\u4e60\u4e0e\u751f\u6210\u5efa\u6a21\uff1a\u4ece\u5355\u4efb\u52a1\u5230\u201c\u52a8\u4f5c\u5206\u5e03\u201d\u7684\u5b66\u4e60<br>\u4f5c\u8005\u5c06\u884c\u4e3a\u514b\u9686\uff08BC\uff09\u7f6e\u4e8e\u201c\u751f\u6210\u6a21\u578b\u201d\u89c6\u89d2\u7cfb\u7edf\u5316\u9610\u8ff0\uff1a<br>1\uff09VAE\uff1a\u4ee5\u6f5c\u53d8\u91cf\u91cd\u5efa\u8f68\u8ff9\uff0c\u9002\u914d\u566a\u58f0\/\u591a\u5cf0\u52a8\u4f5c\u5206\u5e03\uff1b<br>2\uff09\u6269\u6563\u6a21\u578b\uff08Diffusion Models\uff09\uff1a\u5728\u52a8\u4f5c\u7a7a\u95f4\u505a\u53bb\u566a\u751f\u6210\uff0c\u5df2\u6210\u4e3a\u89c6\u89c9\u2014\u8fd0\u52a8\u7b56\u7565\u5b66\u4e60\u7684\u524d\u6cbf\u65b9\u6848\uff1b<br>3\uff09Flow Matching\uff1a\u5728\u8fde\u7eed\u65f6\u95f4\u4e0a\u62df\u5408\u6570\u636e\u5230\u5148\u9a8c\u7684\u6982\u7387\u6d41\uff0c\u63a8\u7406\u66f4\u5feb\u3001\u7a33\u5b9a\u6027\u66f4\u597d\u3002\u8fd9\u4e9b\u751f\u6210\u6cd5\u7684\u5171\u540c\u70b9\u662f\u201c\u5b66\u8f68\u8ff9\u65cf\u201d\uff08\u800c\u975e\u5355\u4e00\u70b9\u63a7\u5236\uff09\uff0c\u5bf9\u591a\u6a21\u6001\/\u4e0d\u786e\u5b9a\u6027\u66f4\u53cb\u597d\u3002\u6587\u4e2d\u968f\u540e\u7ed9\u51fa\u4e24\u6761\u843d\u5730\u4e3b\u7ebf\uff1a<br>A\uff09ACT\uff08Action Chunking with Transformers\uff09\uff1a\u4ee5\u77ed\u65f6\u52a8\u4f5c\u5757\u4e3a\u5efa\u6a21\u5355\u5143\uff0cTransformer \u9884\u6d4b\u672a\u6765\u591a\u6b65\u63a7\u5236\uff0c\u5929\u7136\u5339\u914d\u201c\u7a97\u53e3\u5316\u201d\u6570\u636e\u63a5\u53e3\uff0c\u5de5\u7a0b\u4e0a\u8bad\u7ec3\u2014\u63a8\u7406\u7b80\u6d01\uff1b<br>B\uff09Diffusion Policy\uff1a\u901a\u8fc7\u52a8\u4f5c\u6269\u6563\u5b9e\u73b0\u9c81\u68d2\u5206\u5e03\u62df\u5408\uff0c\u5df2\u5728\u591a\u79cd\u64cd\u4f5c\u4efb\u52a1\u4e0a\u7ed9\u51fa\u5f3a\u6027\u80fd\u4e0e\u6cdb\u5316\u793a\u8303\u3002\u6559\u7a0b\u914d\u6709\u5b8c\u6574\u7684\u8bad\u7ec3\/\u63a8\u7406\u4ee3\u7801\u7247\u6bb5\uff0c\u5f3a\u8c03\u4e0e LeRobotDataset \u7684\u5373\u63d2\u5373\u7528\u3002<\/p>\n\n\n\n<p>\u516d\u3001\u63a8\u7406\u4f18\u5316\uff1a\u8ba9\u7b56\u7565\u201c\u65e2\u5feb\u53c8\u7a33\u201d<br>\u4e3a\u7f29\u77ed\u63a7\u5236\u56de\u8def\u5ef6\u8fdf\u5e76\u63d0\u5347\u5b9e\u65f6\u6027\uff0c\u6559\u7a0b\u63d0\u51fa\u628a\u201c\u52a8\u4f5c\u89c4\u5212\u201d\u4e0e\u201c\u52a8\u4f5c\u6267\u884c\u201d\u89e3\u8026\uff0c\u8f85\u4ee5\u5f02\u6b65\u63a8\u7406\u6808\u4e0e\u6279\u5904\u7406\/\u6d41\u6c34\u7ebf\u5316\uff0c\u5b9e\u73b0\u786c\u4ef6\u4fa7\u7684\u65f6\u5e8f\u4fdd\u969c\u4e0e\u7b56\u7565\u4fa7\u7684\u541e\u5410\u4f18\u5316\uff0c\u5e76\u63d0\u4f9b\u5f02\u6b65\u63a8\u7406\u793a\u4f8b\u4ee5\u4fbf\u5728\u8d44\u6e90\u53d7\u9650\u7684\u771f\u5b9e\u673a\u5668\u4eba\u4e0a\u843d\u5730\u3002<\/p>\n\n\n\n<p>\u4e03\u3001\u901a\u7528\u673a\u5668\u4eba\u7b56\u7565\uff08Generalist Policies\uff09\uff1a\u4ece\u5355\u4efb\u52a1\u5230\u201c\u591a\u4efb\u52a1\u00d7\u591a\u673a\u4f53\u00d7\u8bed\u8a00\u6761\u4ef6\u5316\u201d<br>\u4f5c\u8005\u5c06\u901a\u7528\u7b56\u7565\u5f52\u5165\u201c\u5e7f\u4e49\u7684 BC \u5bb6\u65cf\u201d\uff0c\u56e0\u4e3a\u5b83\u4eec\u672c\u8d28\u4e0a\u4ecd\u4ee5\u5927\u89c4\u6a21\u6f14\u793a\u4e3a\u76d1\u7763\u4fe1\u53f7\uff0c\u53ea\u662f\u5f15\u5165\u8bed\u8a00\/\u56fe\u50cf\u7b49\u591a\u6a21\u6001\u6761\u4ef6\u4e0e\u8de8\u4efb\u52a1\u591a\u673a\u4f53\u6570\u636e\uff1a<br>1\uff09VLA\uff08Vision-Language-Action\uff09\uff1a\u4ee5 VLM\uff08\u5982 PaLM-E\/\u6307\u4ee4\u5fae\u8c03\u67b6\u6784\uff09\u4e3a\u611f\u77e5\u2014\u8bed\u4e49\u5c42\uff0c\u8f93\u51fa\u52a8\u4f5c\u5e8f\u5217\uff0c\u652f\u6301\u8bed\u8a00\u6307\u4ee4\u9a71\u52a8\u7684\u4efb\u52a1\u6cdb\u5316\uff1b<br>2\uff09\u03c00\uff1a\u5f3a\u8c03\u4ece\u5927\u89c4\u6a21\u3001\u8de8\u573a\u666f\u7684\u6f14\u793a\u4e2d\u5b66\u4e60\u7edf\u4e00\u7684\u591a\u4efb\u52a1\u7b56\u7565\uff0c\u914d\u5957\u5f00\u6e90\u63a8\u7406\u4e0e\u8c03\u7528\u793a\u4f8b\uff1b<br>3\uff09SmolVLA\uff1a\u4ee5\u66f4\u201c\u8f7b\u91cf\u201d\u7684\u53c2\u6570\u89c4\u6a21\u8ffd\u6c42\u66f4\u9ad8\u7684\u90e8\u7f72\u6027\u4e0e\u66f4\u4f4e\u5ef6\u8fdf\uff0c\u540c\u6837\u63d0\u4f9b\u4f7f\u7528\u6837\u4f8b\u3002\u6559\u7a0b\u5728\u201c\u6a21\u578b\u4e0e\u6570\u636e\u9884\u5907\u201d\u4e2d\u7ed9\u51fa\u6570\u636e\u914d\u65b9\u4e0e\u8bad\u7ec3\u63a5\u53e3\uff0c\u529b\u56fe\u628a\u201c\u901a\u624d\u673a\u5668\u4eba\u201d\u4ece\u8bba\u6587\u8d70\u5411\u53ef\u590d\u73b0\u5de5\u7a0b\u3002<\/p>\n\n\n\n<p>\u516b\u3001\u65b9\u6cd5\u8bba\u8109\u7edc\u4e0e\u5b9e\u8df5\u5efa\u8bae<br>1\uff09\u8303\u5f0f\u4e92\u8865\uff1a\u5e76\u975e\u201c\u5b66\u4e60\u53d6\u4ee3\u4e00\u5207\u201d\u3002\u5728\u89c4\u5219\u6e05\u6670\u3001\u52a8\u529b\u5b66\u53ef\u51c6\u786e\u5efa\u6a21\u7684\u573a\u666f\uff0c\u7ecf\u5178\u65b9\u6cd5\u7684\u53ef\u89e3\u91ca\u4e0e\u7a33\u5b9a\u4f18\u52bf\u4ecd\u4e0d\u53ef\u66ff\u4ee3\uff1b\u5728\u63a5\u89e6\/\u975e\u7ebf\u6027\/\u591a\u6a21\u6001\u4e0e\u8de8\u4efb\u52a1\u6cdb\u5316\u8bc9\u6c42\u5f3a\u7684\u573a\u666f\uff0c\u5b66\u4e60\u8303\u5f0f\u66f4\u5177\u4f38\u7f29\u6027\u3002<br>2\uff09\u6570\u636e\u4f18\u5148\u4e0e\u6807\u51c6\u5316\uff1a\u4ee5 LeRobotDataset \u4e3a\u67a2\u7ebd\uff0c\u7edf\u4e00\u7279\u5f81\u6a21\u5f0f\/\u7edf\u8ba1\/\u89c6\u9891\u2014\u8868\u683c\u2014\u5143\u6570\u636e\u7684\u7d22\u5f15\u91cd\u5efa\uff0c\u624d\u80fd\u628a\u201c\u5927\u6570\u636e+\u5927\u6a21\u578b\u201d\u7684\u7ea2\u5229\u6765\u5230\u673a\u5668\u4eba\u7684\u201c\u6808\u5e95\u201d\u3002<br>3\uff09\u73b0\u5b9e\u8bad\u7ec3\u4e09\u4ef6\u5957\uff1a\u5b89\u5168\u62a4\u680f\uff08\u9650\u5e45\/\u770b\u95e8\u72d7\/\u6025\u505c\uff09\u3001\u4eba\u7c7b\u5728\u73af\uff08\u5e72\u9884\/\u6f14\u793a\u6ce8\u5165\uff09\u3001\u9ad8\u6548\u7b97\u6cd5\uff08\u79bb\u7b56\u7565+\u91cd\u653e+\u6f14\u793a\u6df7\u5165\/\u5956\u52b1\u5b66\u4e60\uff09\u3002<br>4\uff09\u4ece\u5355\u4efb\u52a1 BC \u5230\u901a\u7528\u7b56\u7565\uff1a\u5148\u7528 ACT\/Diffusion Policy \u5728\u7279\u5b9a\u673a\u4f53\/\u573a\u666f\u6253\u901a\u5de5\u7a0b\u94fe\uff0c\u518d\u5faa\u5e8f\u8fc8\u5411\u8bed\u8a00\u6761\u4ef6\u5316\u7684 VLA\/\u03c00\/SmolVLA\u3002<br>5\uff09\u63a8\u7406\u4fa7\u5de5\u7a0b\uff1a\u5f02\u6b65\u3001\u89e3\u8026\u4e0e\u6d41\u6c34\u7ebf\u5316\u662f\u628a\u201c\u8bba\u6587\u7ea7\u6027\u80fd\u201d\u642c\u5230\u201c\u63a7\u5236\u7ea7\u5b9e\u65f6\u201d\u7684\u5fc5\u8981\u6761\u4ef6\u3002<\/p>\n\n\n\n<p>\u4e5d\u3001\u603b\u7ed3\u4e0e\u5c55\u671b<br>\u6559\u7a0b\u7684\u6838\u5fc3\u4ef7\u503c\u5728\u4e8e\u201c\u6865\u63a5\u201d\uff1a<strong>\u5b83\u4e00\u65b9\u9762\u5c0a\u91cd 60 \u4f59\u5e74\u7ecf\u5178\u673a\u5668\u4eba\u5b66\u7684\u6210\u679c\u4e0e\u8fb9\u754c\u6761\u4ef6\uff0c\u53e6\u4e00\u65b9\u9762\u4ee5\u6570\u636e\/\u6a21\u578b\/\u5de5\u7a0b\u4e09\u4f4d\u4e00\u4f53\u7684\u65b9\u5f0f\uff0c\u628a RL\/BC\/\u751f\u6210\u5efa\u6a21\u4e0e\u901a\u7528\u7b56\u7565\u7684\u6700\u65b0\u8fdb\u5c55\u843d\u5728\u53ef\u590d\u73b0\u7684 lerobot \u4ee3\u7801\u4e0e\u6570\u636e\u6807\u51c6\u4e0a<\/strong>\u3002\u9762\u5411\u672a\u6765\uff0c\u4f5c\u8005\u770b\u597d\u4e09\u6761\u7ebf\uff1a<br>\uff081\uff09\u66f4\u5b89\u5168\u3001\u66f4\u9ad8\u6548\u7684\u771f\u5b9e\u4e16\u754c RL\uff08\u5c11\u6837\u672c\u3001\u4eba\u7c7b\u5728\u73af\u3001\u5956\u52b1\u5b66\u4e60\uff09\uff1b<br>\uff082\uff09\u4ee5\u6269\u6563\/Flow Matching \u4e3a\u6838\u5fc3\u7684\u9c81\u68d2 BC \u4e0e\u5feb\u901f\u63a8\u7406\uff1b<br>\uff083\uff09\u8de8\u4efb\u52a1\/\u8de8\u673a\u4f53\u3001\u8bed\u8a00\u6761\u4ef6\u5316\u7684\u901a\u7528\u7b56\u7565\u4e0e\u5c0f\u578b\u5316\u90e8\u7f72\u8def\u7ebf\u5e76\u8fdb\u3002\u5bf9\u4ea7\u4e1a\u754c\u800c\u8a00\uff0c\u8fd9\u5957\u65b9\u6cd5\u8bba\u7b49\u4e8e\u63d0\u4f9b\u4e86\u4e00\u6761\u4ece\u201c\u5c0f\u578b\u3001\u53ef 3D \u6253\u5370\u7684 SO-100 \u6559\u5b66\/\u539f\u578b\u5e73\u53f0\u201d\u5230\u201c\u901a\u624d\u673a\u5668\u4eba\u201d\u7684\u5de5\u7a0b\u4e0a\u624b\u8def\u5f84\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>LeRobot on GitHub: <a href=\"https:\/\/github.com\/huggingface\/lerobot\" target=\"_blank\" rel=\"noreferrer noopener\">https:\/\/github.com\/huggingface\/lerobot<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Robot Learning: A Tutorial\u662f\u4e00\u7bc7\u9762\u5411\u7814\u7a76\u8005\u4e0e\u5b9e\u8df5\u8005\u7684\u201c\u673a\u5668\u4eba\u5b66\u4e60\u201d\u6559\u7a0b\u578b\u7efc\u8ff0\uff0c\u4e3b\u5f20\u5728 [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"set","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[8],"tags":[39,69,61],"class_list":["post-6366","post","type-post","status-publish","format-standard","hentry","category-tech","tag-ai","tag-engineering","tag-61"],"rttpg_featured_image_url":null,"rttpg_author":{"display_name":"NullThought","author_link":"https:\/\/nullthought.net\/?author=1"},"rttpg_comment":0,"rttpg_category":"<a href=\"https:\/\/nullthought.net\/?cat=8\" rel=\"category\">Tech<\/a>","rttpg_excerpt":"Robot Learning: A Tutorial\u662f\u4e00\u7bc7\u9762\u5411\u7814\u7a76\u8005\u4e0e\u5b9e\u8df5\u8005\u7684\u201c\u673a\u5668\u4eba\u5b66\u4e60\u201d\u6559\u7a0b\u578b\u7efc\u8ff0\uff0c\u4e3b\u5f20\u5728&hellip;","_links":{"self":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/6366","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=6366"}],"version-history":[{"count":1,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/6366\/revisions"}],"predecessor-version":[{"id":6368,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/6366\/revisions\/6368"}],"wp:attachment":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6366"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6366"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6366"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}