摘要
针对仿人机器人学习人类运动的问题,提出一种基于Kinect的仿人机器人的实时模仿系统。首先,利用Kinect设备独有的骨骼追踪功能,实时高精度的完成动作捕捉任务;其次,利用Kinect捕捉到的人体各节点在三维空间中的位置信息进行运动学建模;最后,根据得到的人体关节角度对Nao机器人对应关节进行逆运动学求解,完成仿人机器人实时模仿人运动。实验结果表明,该系统能够实时、准确地完成对目标人物的动作模仿。
Proposes a Kinect-based real-time motion simulation system for humanoid robots to learn human motions. This method uses the skeleton tracking function of Kinect equipment to complete motion capturing. Carries out Kinematic modeling based on the position information of human nodes captured by Kinect in three-dimensional space, then according to the angle of human joints and inverse motion of the joints, solves the problem about simulating the human motion in real time. The experimental results show that the method can simulate the action of target accurately.
作者
孔令轲
KONG Ling-Ke(Fujian Provincial Academic Engineering Research Centre in Industrial Intellectual Techniques and Systems, Engineering College,Huaqiao University, Quanzhou 362021)
出处
《现代计算机》
2019年第17期55-58,共4页
Modern Computer
