测绘通报 ›› 2020, Vol. 0 ›› Issue (6): 7-11,16.doi: 10.13474/j.cnki.11-2246.2020.0171

• 室内定位与导航 • 上一篇    下一篇

融合上下文感知的地标检测辅助WiFi-PDR室内定位

何肖娜1, 宋斌斌1, 余敏2   

  1. 1. 江西师范大学软件学院, 江西 南昌 330022;
    2. 江西师范大学计算机信息工程学院, 江西 南昌 330022
  • 收稿日期:2019-08-02 出版日期:2020-06-25 发布日期:2020-07-01
  • 通讯作者: 余敏。E-mail:myu821@163.com E-mail:myu821@163.com
  • 作者简介:何肖娜(1994-),女,硕士生,主要研究方向为室内定位。E-mail:916835390@qq.com
  • 基金资助:
    国家重点研发计划(2016YFB0502204);国家自然科学基金(41764002)

Context-aware landmark detection assists WiFi-PDR indoor localization

HE Xiaona1, SONG Binbin1, YU Min2   

  1. 1. College of Software, Jiangxi Normal University, Nanchang 330022, China;
    2. College of Computer Information and Engineering, Jiangxi Normal University, Nanchang 330022, China
  • Received:2019-08-02 Online:2020-06-25 Published:2020-07-01

摘要: 针对当前WiFi-PDR室内定位中存在的WiFi信号不稳定及行人航位推算(PDR)累积误差大的问题,本文提出了一种融合上下文感知的地标检测辅助WiFi-PDR室内定位方法。该方法利用智能手机所能监测到的上下文信息建立用户模型,采用基于卷积神经网络的用户行为感知和基于WiFi-PDR室内定位的粗粒度位置感知,发现隐藏的室内地标信息,并完成用户在地标位置的位置校正,提高定位准确度。该方法在一定程度上降低了WiFi-PDR室内定位的误差,提高了用户室内定位的精度。经试验验证,该室内定位方法的精度相比于传统的WiFi-PDR方法提高了43.62%。

关键词: 室内定位, 卷积神经网络, 上下文感知, 行为感知, 位置感知, 地标

Abstract: Aiming at the instability of WiFi signal and the large cumulative error of PDR(pedestrian dead reckoning) in the current WiFi-PDR indoor localization, a context-based landmark detection assisted WiFi-PDR indoor localization method is proposed. This method uses the context information monitored by smart phones to establish a user model, and uses the user behavior perception based on convolutional neural network and coarse-grained location perception based on WiFi-PDR indoor localization to discover the hidden indoor landmark information, and completes the user's location correction at the landmark location to improve the positioning accuracy. This method reduces the indoor localization error of WiFi-PDR to some extent and improves the indoor localization accuracy of users. The experimental results show that the accuracy of the indoor localization method is 43.62% higher than that of the traditional WiFi-PDR method.

Key words: indoor localization, convolutional neural network, context awareness, behavioral perception, location awareness, landmark

中图分类号: