Loading...
|
Please use this identifier to cite or link to this item:
https://nccur.lib.nccu.edu.tw/handle/140.119/134201
|
| Title: | 以手機結合卷積神經網路深度學習實現室內位置追蹤
Smartphone-Based Indoor Position Tracking with CNN Deep Learning |
| Authors: | 吳宛庭
Wu, Wan-Ting |
| Contributors: | 蔡子傑
Tsai, Tzu-Chieh
吳宛庭
Wu, Wan-Ting |
| Keywords: | 深度學習
室內位置追蹤
卷積神經網路
手機感測器
Deep Learning
Indoor Localization
Indoor Position Tracking
Convolutional Neural Network
Smartphone Sensor
CNN
IMU |
| Date: | 2021 |
| Issue Date: | 2021-03-02 14:56:47 (UTC+8) |
| Abstract: | GPS訊號因涵蓋範圍擴及全球，目前被廣泛運用於室外定位，而室內環境由於缺乏訊號覆蓋，無法獲得良好的定位效果。因此過去幾年許多學者致力於各種室內定位研究，包含使用感測器或信號發射裝置，以數據差異或信號強弱辨別使用者所在位置，而這些通常需要事先架設感測器，及量測信號地圖。現今生活中手機方便性極高，若只使用手機內感測器進行室內位置追蹤，是我們認為最便利且直觀的方式。
近幾年「AI人工智慧」技術蓬勃發展，隨著機器學習領域軟硬體逐漸成熟，大幅提高資訊設備的運算速度，更容易以深度學習的框架開發相關應用。我們在本篇論文提出一套深度學習方法，這套方法以手機內的加速度感測器資訊作為圖片資料樣本，並用卷積神經網路模型進行樣本訓練，將此模型部署於手機上後，使用我們提出的平均機制計算出模型的步長預測，搭配手機指南針數據，實現用戶在室內位置的追蹤。我們的實驗讓不同身高的受試者進行測試，實驗的結果具有極高準確度，實驗過程利用我們開發的手機APP在政大建築物內進行完整展示，讓手機用戶在室內環境行走中，達到即時位置追蹤的效果，成果令人相當滿意。
GPS signals are widely used for outdoor localization due to their worldwide coverage, however, may not be applicable for indoor environments to achieve good localization results. Over the past year, there were many indoor localization studies, including the use of various sensors or signal strengths to identify the location of users or devices. These usually require infrastructure and signal measurements in advance. Nowadays, for convenience, if only sensors in smartphones are used for indoor localization, it will be the most intuitive way.
In recent years, "AI artificial intelligence" technology has been developing rapidly. With the increasing computing speed of information appliances, it becomes much easier to develop the related applications with the framework of deep learning. In this thesis, we propose a deep learning method. This method uses the acceleration sensor information in a smartphone as the image data sample, and uses the convolutional neural network model for sample training. We further develop an average mechanism for prediction of step lengths. By combining with compass data from the smartphone, it can track the user`s indoor location. We deploy our model on the smartphone APP which can display the route in real time while we walk. We tested it in the building of NCCU for users of different heights. Experiments results are quite satisfactory with only 1.8% error. |
| Reference: | [1]Paramvir Bahl, & Venkata N. Padmanabhan. (2000, March). RADAR: An in-building RF-based user location and tracking system, Infocom Nineteenth Joint Conference of the IEEE Computer & Communications Societies IEEE, Tel Aviv, Israel.
[2]Magdy Ibrahima, & Osama Moselhib. (2015, June). IMU-Based Indoor Localization for Construction Applications, 32nd International Symposium on Automation and Robotics in Construction, Oulu, Finland. doi:10.22260/ISARC2015/0059
[3]Estefania Munoz Diaz, & Ana Luz Mendiguchia Gonzalez. (2014, October). Step Detector and Step Length Estimator for an Inertial Pocket Navigation System, 2014 International Conference on Indoor Positioning and Indoor Navigation, Busan, South Korea. doi:10.1109/IPIN.2014.7275473
[4]Ahmad Abadleh, Eshraq Al-Hawari, Esra`a Alkafaween, & Hamad Al-Sawalqah. (2017, May). Step Detection Algorithm For Accurate Distance Estimation Using Dynamic Step Length, 2017 18th IEEE International Conference on Mobile Data Management, Daejeon, South Korea. doi:10.1109/MDM.2017.52
[5]Dihong Wu, Ao Peng, Lingxiang Zheng, Zhenyang Wu, Yizhen Wang, Biyu Tang,…Huiru Zheng. (2017, September). A Smartphone Based Hand-Held Indoor Positioning System, 2017 International Conference on Indoor Positioning and Indoor Navigation, Sapporo, Japan. doi:10.1109/IPIN.2017.8115915
[6]Yi-Shan Li, & Fang-Shii Ning. (2018 December). Low-Cost Indoor Positioning Application Based on Map Assistance and Mobile Phone Sensors, Sensors, 18(12), 4285. MDPI AG. Retrieved from http://dx.doi.org/10.3390/s18124285
[7]Baoding Zhou,Jun Yang , & Qingquan Li. (2019). Smartphone-Based Activity Recognition for Indoor Localization Using a Convolutional Neural Network. Sensors, 19(3), 621. MDPI AG. Retrieved from http://dx.doi.org/10.3390/s19030621
[8]Ayush Mittal, Saideep Tiku, & Sudeep Pasricha. (2018). Adapting Convolutional Neural Networks for Indoor Localization with Smart Mobile Devices. GLSVLSI `18: Great Lakes Symposium on VLSI 2018, 117–122. https://doi.org/10.1145/3194554.3194594
[9]Jiheon Kang, Joonbeom Lee, & Doo-Seop Eom. (2018). Smartphone-Based Traveled Distance Estimation Using Individual Walking Patterns for Indoor Localization. Sensors, 18(9), 3149. MDPI AG. Retrieved from http://dx.doi.org/10.3390/s18093149 |
| Description: | 碩士
國立政治大學
資訊科學系碩士在職專班
104971014 |
| Source URI: | http://thesis.lib.nccu.edu.tw/record/#G0104971014 |
| Data Type: | thesis |
| DOI: | 10.6814/NCCU202100345 |
| Appears in Collections: | [資訊科學系碩士在職專班] 學位論文
|
Files in This Item:
| File |
Description |
Size | Format | |
|---|
| 101401.pdf | | 2903Kb | Adobe PDF2 | 0 | View/Open |
|
All items in 政大典藏 are protected by copyright, with all rights reserved.
|
