應用於無人飛行器的室內定位及追蹤系統

Indoor Positioning and Tracking System for Drones

陳柏君(Chen, Po Chune)

指導教授 ： 馬席彬

國立清華大學/電機資訊學院/電機工程學系/碩士(2015年)

摘要

近年來，無人飛行器的相關應用越來越普遍，如攝影、製圖、廣告以及災害搜救。在這篇論文裡，將提出一個應用於無人飛行器的室內定位及追蹤系統。系統分為三個區塊：定位的區塊用來得知特定無線裝置在空間中的位置、追蹤的區塊利用圖形識別來追蹤被選取的物件、移動的區塊決定飛機的移動方向。
我們使用了一款可以限制通訊範圍且可以利用內建微控制器作運算的無線裝置實現了改良的Cell of Origin(CoO)定位演算法。而對於追蹤區塊，我們實現了Tracking-Learning-Detection (TLD)追蹤演算法。TLD是一種可以追蹤位置物件以及學習物件圖樣的計算機視覺演算法。一旦追蹤成功，其誤差距離會小於10公分。移動區塊利用定位以及追蹤的結果來判斷飛機要如何移動，目標是要讓被追蹤的物件維持在鏡頭畫面中央，以及可以自動飛行到使用者指定的室內區域。
最終，我們提出了一個結合CoO定位演算法、TLD追蹤演算法和移動演算法且適用於無人飛行器的系統。此系統使用無線裝置來定位，然後使用無人飛行器的鏡頭來追蹤，最後將鏡頭對準物件且自動移動到指定區域。實驗結果顯示定位的精準度為2公尺、追蹤的成功率高於其他對照演算法33%。除此之外，我們的定位系統主要的優勢為不需為室內環境建立專屬模型，且定位的涵蓋範圍可以擴張。再者，無線裝置的體積為3公分*2公分*2公分，使定位系統擁有可攜式的特性。最後，無線裝置在運作電壓3V的耗能為49.9 mW，在850mAh電池的供應下可以持續運作2天以上。

關鍵字

無人飛行器 ； 室內定位系統 ； 圖形識別

並列摘要

Recently, the drone-related applications are more and more popular such as photography, mapping, real estate and disaster response. In this thesis, the indoor positioning and tracking system for drones is proposed. It includes three parts: positioning component locates position of specified wireless device; tracking component tracks the selected object by pattern recognition; and movement decision component decides the drone’s movement.
We have implemented the modified cell of origin (CoO) indoor positioning algorithm with the wireless device, which is the one limits its communication range and processes some calculation with the built-in microcontroller. For the tracking component, we have implemented tracking-learning-detection (TLD) algorithm. TLD is a computer vision algorithm that tracks unknown objects and recognizes object patterns. If tracking is successful, the margin of error will be less than 10 cm. The movement decision component computes feedback from positioning and tracking components to generate the drone’s movement. It positions objects to the center of image and directs the drone to the location commanded by the user.
Finally, we propose a system which combines the CoO indoor positioning algorithm, the TLD algorithm, and the movement decisions for drones. The system locates the position of drone with the wireless device, tracks objects through drone’s camera, aims drone’s camera to the object, and flies the drone to desired locations. The test results show 2 m accuracy of positioning, and improved successful tracking rate of 33% higher than other tracking algorithms as control groups. The positioning algorithm contributes to the environment model unnecessary and the scalable coverage of positioning. Moreover, the wireless device’s size of
3 cm * 2 cm * 2 cm makes the system portable and the power consumption of 49.9 mW under 3.0 V operation voltage on a 850 mAh battery keeps the wireless device working over 2 days.

並列關鍵字

Drones ； Indoor positioning system ； Pattern recognition

參考文獻

[2] Y. Gu, A. Lo, and I. Niemegeers, “A survey of indoor positioning systems for wireless personal networks,” 2009 IEEE Communications Surveys Tutorials, vol. 11, no. 1, pp. 13–32, First 2009.

[4] M. Al-Ammar, S. Alhadhrami, A. Al-Salman, A. Alarifi, H. Al-Khalifa, A. Alnafessah, and M. Alsaleh, “Comparative survey of indoor positioning technologies, techniques, and algorithms,” in 2014 International Conf. Cyberworlds, Oct. 2014, pp. 245–252.

[5] Z. Song, G. Jiang, and C. Huang, “A survey on indoor positioning technologies,” in Theoretical and Mathematical Foundations of Computer Science. Springer, 2011, pp. 198–206.

[6] J.-g. Liu, D.-M. Shi, and M. K. Leung, “Indoor navigation system based on omni-directional corridorguidelines,” in 2008 IEEE International Conf. Machine Learning and Cybernetics, vol. 3, 2008, pp. 1271–1276.

[8] H. Liu, H. Darabi, P. Banerjee, and J. Liu, “Survey of wireless indoor positioning techniques and systems,” IEEE Trans. Systems, Man, and Cybernetics, Part C: Applications and Reviews, vol. 37, no. 6, pp. 1067–1080, 2007.