以粒子濾波器為基礎的定位系統中利用感測器輔助估測人類行動模型

高齡化社會成為了現今共同的問題，老年人照護的議題也隨著日益重要。養護中心提供了一個老年人安全監控以及集中照護的社群。雙連安養中心也是其中之一，集中照護超過350位長者。安養中心提供了各式的照護服務，不過，最重要的問題仍在老年人的安全。為了監控這麼大數量的長者，人力資源的成本必須花費相當大。我們應用了以RSSI(接收信號強度)為基礎的定位系統來幫助安全監控的問題。定位的演算法是以KNN估測為基礎，加上粒子濾波器使得定位點輸出流暢化。但KNN估測很容易被無線信號強度的不穩定性所影響，這會使得定位精確度大幅下降。粒子濾波器便是用以解決這個問題。但戶外環境和室內環境是有很大的不同。在室內環境裡，人的移動受限於有限的空間限制。所以在粒子濾波器中的移動模型，只要設定在一個平均的移動率即足以克服室內環境裡人的各種行動模式。然而，人們再是外會有各種不同的行動模式，譬如跑步、慢跑、走路等。在室外的環境中很容易會有比較極端的移動模式。若將行動模型設定在一個平均值上恐怕不足以解決粒子濾波器中追蹤的問題。這篇論文找出了一個利用附加加速度器的方法來收集加速度值，並且利用利用FFT(快速傅立葉轉換)來分析資料。藉此，我們可以找出監控目標的步頻，乘上目標的步距，即可線上估測目標速度。利用這個速度的估計值，粒子濾波器也可以同時調整定位估測所使用的行動模型，進而提升定位精確率。

關鍵字

KNN估測；步頻； FFT(快速傅立葉轉換) ；行動模型

並列摘要

Nowadays, aging population becomes a common problem in the world. Elderly caring is a getting emphasized issue. Nursing center turns out to be a community for elderly people to centralize safety monitoring and caring. Suang-lien nursing center is one of them. There are over 350 members reside in it. The nursing center provides a wide variety of caring services for elderly people, hence the most important problem is their safety concern. In order to monitoring large amount of the elderly people, human resource on the caring is costly. We apply our RSSI based localization system to help safety monitoring. Our localization algorithm is based on KNN estimation plus particle filter to smooth position output. The KNN estimation is easily affected by wireless signal instability, which devastates the location accuracy. Thus the particle filter is used to solve the issue. Outdoor environment is much different than indoor. In indoor environment, human moves within the restriction of limited space. Therefore, set the mobility model inside particle filter with average mobility is enough to solve human moving patterns in indoor environment. However, human might have various mobility patterns in outdoor, including running, jogging, walking…etc. Extreme mobility cases are easily to occur in outside environment. Set the mobility in average case is not enough to solve the tracking problem. Our work comes up an idea that we could use an extra added accelerometer and analyze acceleration values by FFT. After that, we could get the target stride frequency, and multiply with target’s stride length. The target speed can be approximated on-line. With the speed approximation, the particle filter can simultaneously adjust the mobility model for position estimation, therefore, enhance the location estimation accuracy.

並列關鍵字

KNN estimation ； stride frequency ； FFT ； mobility model

參考文獻

[1] N. Bulusu, J. Heidemann, and D. Estrin, “GPS-less low-cost outdoor localization for very small devices,” Personal Communications, IEEE, vol. 7, no. 5, pp. 28–34, 2000.

[3] A. Smailagic and D. Kogan, “Location sensing and privacy in a context-aware computing environment,” IEEE Wireless Communications, vol. 9, no. 5, pp. 10–17, 2002.

[4] P. Bahl and V. N. Padmanabhan, “RADAR: an in-building RF-based user location and tracking system”, in Proceedings of the 19th Annual Joint Conference of IEEE Computer and Communications Societies (INFOCOM ’00), vol. 2, pp. 775–784, Tel Aviv, Israel, March 2000.

[5] K. Lorincz and M. Welsh, “MoteTrack: A Robust, Decentralized Approach to RF-Based Location Tracking,” in Proceedings of the International Workshop on Location and Context-Awareness (LoCA), pp. 63-82, May 2005.

[6] Vinay Seshadri, Gergely V. Z’aruba, and Manfred Huber, “A Bayesian Sampling Approach to In-Door Localization of Wireless Devices Using Received Signal Strength Indication”, in PerCom, pages 75–84, 2005.

國際替代計量

以粒子濾波器為基礎的定位系統中利用感測器輔助估測人類行動模型

全文下載

主題瀏覽