基于数字技术和负载均衡的AI体育教育机器人智能教学模型构建  

Construction of intelligent teaching model of AI physical education robot based on digital technology and load balancing

作　　者：刘亮[1] LIU Liang(Xi’an Aeronautical Polytechnic Institute,Xi’an 710089,China)

机构地区：[1]西安航空职业技术学院,西安710089

出　　处：《自动化与仪器仪表》2025年第3期177-181,共5页Automation & Instrumentation

基　　金：陕西省教育厅教改课题《基于学生职业能力发展的高职院校网络通识课程应用与管理研究》(17GY008)。

摘　　要：随着数字化教育的不断深入,人工智能快速渗入教育领域。研究旨在融合深度学习算法和硬件资源优化方法,建立一种智能教学模型。研究结合目标检测、骨骼检测算法和图卷积神经网络完成人体动作识别,采用动态时间规整算法进行评分,并引入负载均衡算法进行优化。对研究提出目标检测算法进行测试发现,目标检测的F1得分为98.7%,均值平均精度为95.2%,优于对比算法。骨骼关键点检测算法的测试结果显示,检测准确率为97.2%,优于对比算法。模型实验结果表明,研究模型初级和高级跆拳道技术动作中的评分准确率分别为97.89%、95.34%,均优于其他模型;模型的评分接近实时状态,平均延迟约9 ms,大幅低于其他模型。该智能教学模型对跆拳道初学者动作提升具有极大的帮助作用。With the deepening of digital education,artificial intelligence is rapidly penetrating into the field of education.The research aims to integrate deep learning algorithm and hardware resource optimization method to establish intelligent teaching model of AI Taekwondo education robot.The study improved Yolov5s,HRNet and graph convolutional neural network,combined with dynamic time normalization algorithm to score students,actions in real time,and adopted dynamic weight load balancing algorithm to optimize resources.The results show that the scoring accuracy of the model in elementary and advanced taekwondo movements is 97.89% and 95.34%,respectively,with an average delay of about 9 ms.The intelligent teaching model is of great help to the movement improvement of Taekwondo beginners.

关 键 词：Yolov5s HRNet GCN DTW 负载均衡 

分 类 号：G87[文化科学—体育训练]