Mana: Dexterous Manipulation of Articulated Tools
中文标题: Mana：关节工具的灵巧操作
英文摘要
Mana is a sim-to-real framework that reinterprets dexterous manipulation of articulated tools as an animation problem. It uses a coarse-to-fine pipeline combining procedurally generated grasp keyframes with motion planning and reinforcement learning. Data generation requires only a few mouse clicks to specify functional affordances, taking less than one minute per tool. The method achieves zero-shot sim-to-real transfer for both grasping and in-hand manipulation across four articulated tools of different scales and joint types. This demonstrates a scalable approach to a challenging robotics problem.
中文摘要
Mana 是一个从仿真到现实（sim-to-real）的框架，将关节工具的灵巧操作重新定义为动画问题。它采用由粗到精的流水线，将程序生成的抓取关键帧与运动规划和强化学习相结合。数据生成仅需几次鼠标点击来指定功能可供性，每个工具不到一分钟。该方法在四种不同规模和关节类型的关节工具上实现了抓取和手中操作的零样本 sim-to-real 迁移，展示了一种可扩展的关节工具使用方案。
关键要点
Reinterprets dexterous manipulation as an animation problem, using a coarse-to-fine pipeline with keyframes, motion planning, and RL.
将灵巧操作重新定义为动画问题，采用由粗到精的流水线，结合关键帧、运动规划和强化学习。
Achieves zero-shot sim-to-real transfer for grasping and in-hand manipulation of articulated tools.
实现了关节工具的抓取和手中操作的零样本 sim-to-real 迁移。
Requires minimal human annotation: a few mouse clicks to specify affordances, under 1 minute per tool.
仅需极少人工标注：几次鼠标点击指定可供性，每个工具不到一分钟。
Demonstrated on four diverse articulated tools, showing scalability across scales and joint types.
在四种不同的关节工具上验证，展示了在不同规模和关节类型上的可扩展性。