摘要
本研究的机器人系统采用计算机视觉对人体姿态进行动态标识,从现有的单目相机获得图像或视频,利用Mediapipe方法对32个人体关键部位进行了定位和可信度的测量。控制系统通过反馈关节位置标记进行人机关节映射,控制驱动元件复现人体实时自然交互,实现人体跟随的功能。该系统的随机实验结果证明了系统设计的正确性和可行性,这对人体姿态跟随机器人系统的仿真研究有一定的指导作用。
The robotic system in this study uses computer vision to dynamically mark human posture,obtains images or videos from existing monocular cameras,and uses the Mediapipe method to localize and plausibly measure 32 key human body parts.The control system performs human body joint mapping by feedback joint position markers and controls the drive elements to reproduce the real time natural interaction of the human body and the function of real human body following.The stochastic experimental results of the system prove the correctness and feasibility of the system design,which is a guideline for the simulation study of human posture following robot system.
作者
张华锐
赵潇帆
王苑丞
吕行
彭熙
ZHANG Hua-rui;ZHAO Xiao-fan;WANG Yuan-cheng;LU Xing;PENG Xi(School of Computer Science,Central China Normal University,Wuhan 430079,China;Key Laboratory of Artificial Intelligence and Intelligent Learning of Hubei Province,Wuhan 430079,China;Demonstration Center of Computer Basic Experimental Teaching of Hubei Province,Wuhan 430079,China)
出处
《电脑与信息技术》
2024年第2期14-17,102,共5页
Computer and Information Technology
基金
华中师范大学大学生创新创业训练计划项目(项目编号:202210511022)
华中师范大学大学生创新创业训练计划项目(项目编号:202210511023)。
