关于实时步长匹配行人室内定位方法①.PDFVIP

计 算 机 系 统 应 用 2017 年 第 26 卷 第 4 期 基于实时步长匹配的行人室内定位方法① 1,2 1 1 1 钟立扬 , 崔超远 , 王儒敬 , 孙丙宇 1( 中国科学院 合肥智能机械研究所, 合肥 230031) 2( 中国科学院大学, 北京 100049) 摘 要: 针对室内定位研究中行人航位推算(PDR)算法的传统步长模型无法充分反映个体差异性的问题, 提出一 种以卫星测距为基础, 建立个体步长特征数据库代替步长模型的方法. 在室外将卫星测距应用到步长测算中, 建 立不同速度特征与步长的对应关系. 在室内, 改进了 K-最近邻(KNN)算法, 对跨步进行实时步长匹配, 结合方向 信息计算出行人的当前位置. 实验结果表明, 室内定位最大误差距离可以控制在 4% 以内, 平均绝对误差距离可 以控制在2% 以内, 定位精度较传统步长模型有较大提高. 关键词: 室内定位; 行人航位推算; 卫星测距; K-最近邻; 实时步长匹配 Method of Pedestrian Indoor Positioning Based on Real-Time Step-Size Matching 1,2 1 1 1 ZHONG Li-Yang , CUI Chao-Yuan ,WANG Ru-Jing , SUN Bing-Yu 1(Institute of Intelligent Machine, Chinese Academy of Sciences, Hefei 230031, China) 2(University of Chinese Academy of Sciences, Beijing 100049, China) Abstract: Focus on the problem of the traditional step-size model used in pedestrian dead reckoning algorithm cannot fully reflect individual differences in the research of indoor positioning. We propose a method to establish database of individual feature of step-size based on satellite ranging and then replace the step-size model. Satellite ranging is applied to calculate the step-size in the outdoor, the corresponding relationship between different speed features and step-size is set up. Then we carry out real-time step-size matching indoors for the step, using improved K-nearest-neighbor(KNN) algorithm and calculate people’s current position combined with direction information. The actual test results show that the maximum error distance can be controlled within 4%, and the mean absolute error distance can be controlled within 2%. The positional accuracy has a larger enhancement compared to the traditional step-size model. Key words: indoor positioning; pedestrian