應用於2.5D環境以粒子過濾演算法與感測元件輔助的無線定位系統

Wireless Location Tracking by a Sensor-Enhanced Particle Filter in 2.5D Buildings

朱瑞浩(Jui-Hao Chu)

指導教授 ： 曾煜棋 ； 易志偉

國立交通大學/資訊學院/資訊科學與工程研究所/碩士(2009年)

https://doi.org/10.6842/NCTU.2009.00741

摘要

在室外空曠環境，GPS具有不錯的定位結果，目前，如何得到使用者在室內環境的位置，受到廣泛的討論和研究，其中尤其以基於無線電訊號強度的樣本比對演算法具有最好的定位結果，但是其最大的缺點在於，無線電訊號先天上容易受到環境影響，導致訊號強度飄移的問題，造成定位結果產生誤差，同時，過去的研究，鮮少討論多層樓等室內立體定位環境 (2.5D環境)，因此，在此篇，我們為無線電訊號強度樣本比對演算法提出了應用於2.5D定位環境的SEPF (sensor-enhanced particle filter) 模型，透過慣性感測元件 (IMU sensors) 感測使用者的移動軌跡，藉以調整粒子 (particles) 的位置分佈和權重值 (weight)。在此篇中，我們會介紹如何建構室內定位環境的2.5D模型，並且在此模型中，粒子的實際取樣 (sampling) 與重新取樣 (re-sampling) 的實作方式，以及藉由分析慣性感測元件的感測值來估測使用者目前的移動行為，例如行走於路面上或樓梯、搭乘電梯等，透過慣性感測元件來克服無線電訊號強度飄移的問題。最後利用我們開發的定位系統雛形，驗證我們的系統效能。

關鍵字

慣性感測元件 ； 定位系統 ； 粒子演算法 ； 遍佈計算 ； 感測網路

並列摘要

For outdoor localization, GPS already provides a satisfactory solution. For indoor localization, however, a globally usable solution is still missing. One promising direction is the pattern-matching solution that relies on RF signals from existing network infrastructures. One major drawback of such systems is the signal-drifting problem, which is an inherent physical constraint. Also, most works only consider single-floor buildings. However, buildings normally have multiple floors (i.e., of 2.5 dimensions). This paper proposes a SEPF (sensor-enhanced particle filter)} model for RF-based pattern-matching localization in a 2.5-D building. IMU sensors are adopted to capture human mobility, while particles reflect the belief on where the user is located. Our framework addresses the following important issues. First, our 2.5-D building model considers multiple floors connected by stairs and elevators. Second, we show how particles should be sampled/re-sampled in a 2.5-D building to reflect change of brief. Third, IMU sensor inputs are exploited to conquer the signal-drifting problem and to predict user's behaviors (walking on grounds/stairs and taking elevators). A prototype has been developed and intensively tested to verify the model.

並列關鍵字

IMU (inertial measurement unit) ； location tracking ； particle filter ； pervasive computing ； sensor network

參考文獻

and A. Hopper. Implementing a Sentient Computing System. Computer,

[3] S. Beauregard,Widyawan, and M. Klepal. Indoor pdr performance enhancement

and Navigation Symposium, 2008 IEEE/ION, pages 141–147, May

[4] E. Brassart, C. Pegard, and M. Mouaddib. Localization using infrared beacons.

Robotica, 18(2):153–161, 2000.