基于动作捕捉的人形机器人遥操作动作重定向技术  

An Optimal Motion Retargeting Approach for Teleoperate Humanoid Robot Based On the Motion Capture

作　　者：王文杰[1] 秦现生[1] 王鸿博[1] 洪杰[1] 牛军龙[1] 谭小群[1] 张雪峰[2] 

机构地区：[1]西北工业大学机电学院,西安710072 [2]中国运载火箭技术研究院总体设计部,北京100076

出　　处：《机械设计与研究》2017年第1期43-49,共7页Machine Design And Research

基　　金：国家自然科学基金资助项目(51275413)

摘　　要：遥操作是机器人研究的重要应用,针对人形机器人开发了一种机器人遥操作方法,利用基于误差最优化的动作重定向技术将人体动作捕捉数据映射到人形机器人模仿人体动作完成遥操作任务,同时能够在遥操作过程在质心调节和三点支撑法自动切换保持平衡。该方法已经在人形机器人NAO进行验证,成功的遥操作NAO机器人应用于书法书写。实验结果显示,人和机器人之间的动作映射保持了良好的运动学匹配和较高的跟踪精度,同时保证了机器人的平衡。Teleoperation is a promising application of humanoid robots. In this paper, we introduce a humanoid- based teleoperation method, that use the motion retargeting technology based on the minimizing the tracking error of the end effectors to map the human mocap data into the robotics for operating, while maintaining balance using a technique that automatically switches the balance control mode between the Center of Mass （ COM ） and three-points support method. Our proposed method has been verified in experiments in a NAO robot, which have successfully completed the teleoperation task in writing Chinese traditional calligraphy. Experimental results show that the motion mapping between the operator and robot have a good kinematic match and small tracking error, while maintaining the balance of the robot.

关 键 词：人形机器人 遥操作 动作重定向 平衡控制 书法 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]