UniGraspTransformer：用于可扩展灵巧机器人抓取的简化策略蒸馏
UniGraspTransformer: Simplified Policy Distillation for Scalable Dexterous Robotic Grasping
AI总结：
提出UniGraspTransformer，通过简化的策略蒸馏流程，将强化学习生成的单物体抓取轨迹蒸馏至单一通用Transformer网络，实现灵巧抓取的高效扩展与泛化，成功率显著优于现有方法。
摘要：
我们引入了 UniGraspTransformer，这是一个用于灵巧机器人抓取的通用 Transformer 网络，它简化了训练同时增强了可扩展性和性能。与 UniDexGrasp++ 等需要复杂多步训练流程的先前方法不同，UniGraspTransformer 遵循一个简化的流程：首先，使用强化学习为单个物体训练专用策略网络以生成成功的抓取轨迹；然后，将这些轨迹蒸馏到一个单一的通用网络中。我们的方法使 UniGraspTransformer 能够有效扩展，包含多达 12 个自注意力块以处理具有不同姿态的数千个物体。此外，它在基于状态和基于视觉的环境中评估时，对理想化和现实世界的输入都能很好地泛化。值得注意的是，UniGraspTransformer 为各种形状和方向的物体生成更广泛的抓取姿态，从而产生更多样化的抓取策略。实验结果表明，与最先进的 UniDexGrasp++ 相比，在各个物体类别上取得了显著改进，在基于视觉的环境中，分别在已见物体、已见类别中未见物体和完全未见物体上实现了 3.5%、7.7% 和 10.1% 的成功率提升。项目页面：https://dexhand.github.io/UniGraspTransformer。
英文摘要：
We introduce UniGraspTransformer, a universal Transformer-based network for dexterous robotic grasping that simplifies training while enhancing scalability and performance. Unlike prior methods such as UniDexGrasp++, which require complex, multi-step training pipelines, UniGraspTransformer follows a streamlined process: first, dedicated policy networks are trained for individual objects using reinforcement learning to generate successful grasp trajectories; then, these trajectories are distilled into a single, universal network. Our approach enables UniGraspTransformer to scale effectively, incorporating up to 12 self-attention blocks for handling thousands of objects with diverse poses. Additionally, it generalizes well to both idealized and real-world inputs, evaluated in state-based and vision-based settings. Notably, UniGraspTransformer generates a broader range of grasping poses for objects in various shapes and orientations, resulting in more diverse grasp strategies. Experimental results demonstrate significant improvements over state-of-the-art, UniDexGrasp++, across various object categories, achieving success rate gains of 3.5%, 7.7%, and 10.1% on seen objects, unseen objects within seen categories, and completely unseen objects, respectively, in the vision-based setting. Project page: https://dexhand.github.io/UniGraspTransformer.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
