人-机器人交互场景中的身份识别技术  

SURVEY ON HUMAN RECOGNITION WITH HUMAN-ROBOT INTERACTION

作　　者：胡荣林 董甜甜 Hu Ronglin;Dong Tiantian(Faculty of Computer and Software Engineering,Huaiyin Institute of Technology,Huaian 223003,Jiangsu,China)

机构地区：[1]淮阴工学院计算机与软件工程学院,江苏淮安223003

出　　处：《计算机应用与软件》2023年第11期1-10,28,共11页Computer Applications and Software

基　　金：国家自然科学基金项目(61603146)。

摘　　要：身份识别是一种使用人的图像、语音等信息,借助目标检测、信息检索等找到其对应标签的综合方法。但是传统身份识别方法都是基于固定终端进行的,无法应对交互环境变化导致的多种数据质量问题,使得基于移动机器人的身份识别任务比基于固定终端的身份识别任务增加了难度。主动视觉、行为预测、视图规划、多模态身份识别等技术的融合运用为此提供了新的解决思路。通过对上述技术的研究进展及应用现状进行综述,分析这些方法在身份识别任务上的优势与不足,并对人-机器人交互场景中的身份识别技术的未来发展趋势进行展望。Identity recognition is a comprehensive method that uses people's image,voice and other information to find their corresponding tags with the help of target detection and information retrieval.However,traditional identification methods are all based on fixed terminals and cannot cope with various data quality problems caused by changes in the interactive environment,which makes the identification task based on mobile robots more difficult than the identification task based on fixed terminals.The fusion of technologies such as active vision,behavior prediction,view planning,and multimodal identification provides new solutions for this.Through the review of the research progress and application status of the above technologies,the advantages and disadvantages of these methods in the identification task are analyzed,and the future development trend of the identification technology in the human-robot interaction scene is prospected.

关 键 词：人-机器人交互 身份识别 主动视觉 多模态身份识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]