A Wearable Interface for Robot Teaching

这项研究旨在开发一种可穿戴界面，直观地教授服务机器人日常任务。该界面测量包括手部动作在内的人类动作，并将其转换为机器人的动作命令。我们首先开发了以下两种方法：(1)一种仅通过附着在人体上的相机和惯性传感器快速可靠地估计身体/手运动的方法；(2)一种基于fsm的状态转换模型和测量的身体/手运动使机器人执行包括操作和运动在内的任务的方法。这两种方法已经在一个实际的人形机器人上进行了测试。我们还对远程机器人教学和人机协作任务进行了初步研究。

This research aims at developing a wearable interface for intuitively teaching everyday tasks to service robots. The interface measures human motions including hand motions and converts them into robot motion commands. We first developed the following two methods: (1) a method of estimating body/hand motions fast and reliably only from a camera and inertial sensors attached to the human body, (2) a method of making the robot perform tasks including manipulation and movement based on an FSM-based state transition model and the measured body/hand motions. These two methods have been tested using an actual humanoid robot. We also have conducted preliminary researches on robot teaching at a remote site and human-robot collaborative tasks.