基于数据手套的多模态手势交互无人机操控方法研究  

UAV Manipulation Methods through Data Glove Based Multimodal Gesture Interaction

作　　者：蒋方园 周方 Jiang Fangyuan;Zhou Fang(Wuhan Railway Vocational College of Technology,Wuhan 430205,China)

机构地区：[1]武汉铁路职业技术学院,湖北武汉430205

出　　处：《湖北成人教育学院学报》2025年第2期119-124,共6页Journal of Hubei Adult Education Institute

基　　金：2023年度湖北省教育厅科学研究计划指导性项目“基于多模态手势交互的无人机操作方法研究”(B2023506)。

摘　　要：随着智能无人机在工业巡检、灾后搜救等领域的广泛应用,传统的操控方式逐渐暴露出效率与灵活性不足的问题。因此研究中提出一种改进型多模态融合算法并引入注意力机制,优化模态特征提取与融合能力。实验结果显示,改进算法在手势识别任务中,准确率达到96.4%,较传统机器学习算法提升超过10%。在模态失效情况下,其鲁棒性显著增强,400次迭代后准确率达到97.8%。无人机操控实验表明,在静态、动态及干扰环境下,操控精度分别为96.8%、94.5%和93.2%,路径偏差最低可达0.18 m,显著优于传统方法。该方法可以显著提高无人机操控的实时性、精度与适应性,为多模态交互技术在智能无人机领域的应用提供有力支撑。With the wide application of intelligent UAVs in industrial inspection,post-disaster search and rescue and other fields,the traditional manoeuvring methods gradually reveal the problems of insufficient efficiency and flexibility.Therefore,in this study,an improved multimodal fusion algorithm and the introduction of an attention mechanism are proposed to optimise the modal feature extraction and fusion capability.Experimental results show that the improved algorithm achieves an accuracy of 96.4%in the gesture recognition task,which is more than 10%higher than the traditional machine learning algorithm.In the case of modal failure,its robustness is significantly improved,reaching an accuracy of 97.8%after 400 iterations.UAV manoeuvring experiments show that the manoeuvring accuracies are 96.8%,94.5%and 93.2%under static,dynamic and interference environments,respectively,and the path deviation can be as low as 0.18m,which is significantly better than the traditional method.This method can significantly improve the real-time,accuracy and adaptability of UAV manipulation,and strongly support the application of multimodal interaction technology in the field of intelligent UAVs.

关 键 词：多模态融合 手势识别 无人机操控 深度学习 路径规划 

分 类 号：G642[文化科学—高等教育学]