Space robots in the space station can assist or substitute astronauts to conduct some in-cabin tasks to alleviate the problems of heavy tasks and the limited number of astronauts in the space station. In response to these problems, a hand gesture teleoperation method is proposed for controlling of dexterous manipulators in the space station, which is based on monocular hand motion capture and video-calling tools. This method can realize free and convenient teleoperation. The work is mainly divided into two parts, which are monocular hand motion capture and teleoperation framework for dexterous manipulators. For monocular hand motion capture, a hand pose estimation method based on hand biological constraints from single RGB images is designed. On the other hand, a dexterous manipulator teleoperation framework based on hand motion capture and video-calling tools is designed. Experimental results carried out on hand pose estimation datasets proved that the proposed hand pose estimation method outperforms the state-of-the-art methods. Physical experiment on a space dexterous manipulator teleoperation platform validated the effectiveness of the proposed hand gesture teleoperation method.

空间站空间机器人可以辅助或替代宇航员进行部分舱内任务，缓解空间站任务繁重、宇航员人数有限等问题。针对这些问题，提出了一种基于单目手部动作捕捉和视频通话工具的空间站灵巧机械臂手势遥操作控制方法。这种方法可以实现自由方便的遥操作。工作主要分为单目手部动作捕捉和灵巧机械手遥操作框架两部分。针对单目手部动作捕捉，设计了一种基于单幅 RGB 图像手部生物约束的手部姿态估计方法。另一方面，设计了一种基于手部动作捕捉和视频通话工具的灵巧机械臂遥操作框架。在手部姿势估计数据集上进行的实验结果证明，所提出的手部姿势估计方法优于最先进的方法。在空间灵巧机械臂遥操作平台上进行的物理实验验证了所提手势遥操作方法的有效性。