Abstract: In order to better promote the robot’s real-time accurate imitation of human movements, a kinematic mapping algorithm based on equivalent joints is proposed for anthropomorphic robotic arms based on the motion characteristics of human arms. This method equates anthropomorphic robotic arm joints to human arm joints based on the characteristics of human arm joint motion, and establishes the conversion model of human arm to robotic arm. And the mapping function is determined by the transformation model and the human arm characteristic function, which solves the rotation axis difference between the human arm joint and the anthropomorphic robotic arm joint. A new inverse kinematics solution method is proposed for the complex problem of inverse kinematics of anthropomorphic robotic arm. The method takes advantage of the feature that the equivalent joint is a spherical joint, and decomposes the anthropomorphic robotic arm inverse kinematics problem into three problems, solving each joint angle separately. Simulation and practical experimental results show that the proposed mapping method can make the robot reproduce human arm movements smoothly and accurately with high end accuracy and perfect bit shape similarity. It is also compared with the joint space-based mapping method and the task space-based mapping method.

Key words: robotic imitation, kinematics mapping, motion similarity, anthropomorphic robotic arm

摘要： 为了更好地促进机器人对人体动作的实时精准模仿，以人臂运动特性为基础，针对拟人机械臂提出了一种基于等效关节的运动学映射算法。该方法根据人臂关节运动特点，将拟人机械臂关节等效为人臂关节，建立人臂到机械臂的转换模型。并由转换模型与人臂特征函数确定映射函数，解决了人臂关节与拟人机械臂关节之间旋转轴线差异问题。针对拟人机械臂逆运动学复杂问题，提出一种新的逆运动学求解方法。该方法利用等效关节为球关节特点，将拟人机械臂逆运动学问题分解成三个问题，分别求解各关节角。仿真和实际实验结果表明，所提出的映射方法能够使机器人平滑且准确地复现人臂动作，拥有较高末端精度与完美的位形相似度。并与基于关节空间的映射算法和任务空间的映射方法进行了比较。

关键词: 机器人模仿, 运动学映射, 运动相似性, 拟人机械臂