面向手势动作捕捉的传感器设计及主从手运动映射

本文针对人手和机器人灵巧手在运动学上尺寸的不一致现象带来的主从手指尖运动空间映射的不一致性进行研究.基于聚氨酯拉伸应变传感器构建了测量手指关节弯曲角度的数据手套.通过建立主从手的运动学模型,基于旋转矩阵理论及正向运动学提出了一种指尖运动轨迹计算方法及手势动作捕捉算法.基于正向运动学及逆向运动学的指尖运动映射算法建立了主从手的指尖运动空间轮廓.建立虚拟实验场景,分别针对关节角度映射算法、手势动作捕捉算法及指尖运动映射算法进行了一系列试验.通过实验得出基于聚氨酯的拉伸应变传感器具有良好的时间响应特性及电学稳定性,基于手势动作捕捉算法能够获取主从手的指尖运动空间轮廓,指尖运动轨迹的计算误差控制在2.8 mm以内.结果证明了基于手势动作捕捉算法以及指尖运动映射算法能够实现主从手指尖运动空间的一致性.

Sensor Design for Hand Gesture Capturing and Master-Slave Hand Motion Mapping

This paper focuses on the inconsistency in motion spatial mapping of the master-slave fingertip caused by the inconsistent dimensions phenomenon of kinematics between human hand and robot dexterous hand. A data glove based on polyurethane strain sensors is made as data interface to measure the joint angles of the fingers. A fingertip motion trajectory calculation method and a hand gestures capturing algorithm are proposed based on rotation matrix theory and forward kinematics by establishing a master-slave kinematic model. The motion workspace boundaries of the master-slave fingertip are established by a fingertip motion mapping algorithm based on forward and inverse kinematics. A virtual experiment platform is established, and a series of experiments are performed for the joint angle mapping algorithm, the gesture motion capturing algorithm, and the fingertip motion mapping algorithm. The experiments show that the strain sensor based on polyurethane has good time response characteristic and electrical stability. The motion workspace boundaries of the master-slave fingertip are acquired based on the gesture motion capturing algorithm, and the calculation error of the fingertip motion trajectory is controlled within 2.8 mm. The results show that the combination of the hand gesture motion capturing algorithm and the fingertips motion mapping algorithm can achieve the consistency of master-slave fingertip motion space.