摘要
针对近年来防疫工作的需要,各种疫情防控机器人快速发展,传统疫情防控机器人的功能多为隔离病房护理和医疗物品递送,而较少能对病人进行核酸采样,本文设计了一种基于树莓派的智能咽拭子采样机器人系统,该系统具有核酸采样、数据上传、人脸识别等功能。当机械臂进行核酸采样时,先使用双目摄像机测距,再通过深度学习训练出可以进行人的面部和口部识别的模型,得到口部三维坐标后,机械臂采用运动学逆解得出末端运动轨迹,自动进行核酸采样。与现有疫情防控机器人相比,本系统填补了在核酸采样方面的空白,具有高效、智能、便捷、灵活的特点。实验测试表明,所设计的核酸采样机器人可以完成口腔位置定位、取咽拭子、口腔咽拭子核酸采样和取放咽拭子等功能,且目标检测的精度达到85%以上,能精确识别不同位置和不同明暗程度的人脸。
In response to the recent need for epidemic prevention,various robots have rapidly developed.Traditional epidemic prevention and control robots mainly perform tasks such as nursing in isolation wards and delivering medical supplies,with less focus on collecting nucleic acid samples from patients.This paper designs an intelligent throat swab sampling robot system based on Raspberry Pi to address this gap.The system integrates functions including nucleic acid sampling,data uploading,and facial recognition.During nucleic acid sampling,the robotic arm first measures distances using binocular cameras,then employs a deep learning model trained for facial and oral recognition.After obtaining the three-dimensional coordinates of the oral cavity,the robotic arm uses kinematic inverse solutions to derive the end-effector trajectory for automated nucleic acid sampling.Compared to existing epidemic prevention and control robots,this system fills the gap in nucleic acid sampling,offering efficiency,intelligence,convenience,and flexibility.Experimental tests demonstrate that the designed nucleic acid sampling robot can accurately locate the oral cavity,obtain throat swabs,perform nucleic acid sampling from oral swabs,and handle swab retrieval and placement.The system achieves an accuracy of over 85%in target detection,accurately identifying faces in different positions and lighting conditions.
作者
王子铭
孙永俣
郑智康
乐洋
WANG Ziming;SUN Yongyu;ZHENG Zhikang;LE Yang(School of Integrated Circuit Science and Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210000,China;School of Computer Science,NJUPT,Nanjing University of Posts and Telecommunications,Nanjing 210000,China;School of Geographic and Biologic Information,Nanjing University of Posts and Telecommunications,Nanjing 210000,China)
出处
《智能计算机与应用》
2024年第8期184-190,共7页
Intelligent Computer and Applications
关键词
深度学习
人脸识别
运动学逆解
deep learning
face identification
inverse kinematics