對於近年來逐漸增多的高齡者與身心障礙者，市面上已有許多行動輔具已經非常成熟，但是對於重度身障者來說，這些行動輔具對於他們仍然有相對高的門檻，針對能力較缺乏的患者，傳統的搖桿控制輪椅、聲控輪椅皆無法合適的讓重度身障者使用，也減少了重度身障者出外活動或者與家人聯繫感情的機會，對於而這些身障者依舊重度依賴著照護者，因此提出一個結合影像物件追蹤與光達避障的互動系統，本系統用於居家可增進臥床的患者與家人互動的情況，適用於外出時降低照護者的負擔，也提供其行車安全性。
本系統架構由嵌入式系統、Android平板電腦、光達測距模組、魚眼廣角相機組成，可適用於居家的影像追蹤小車，以及外出的電動輪椅系統，使用魚眼相機可增加視角幫助對周圍敏感度低的重障者，另外以容易取得的平板電腦作為使用者介面，平板電腦上的Android APP顯示行車過程所需的安全資訊，而為了任務的分工，兩塊嵌入式系統微控制器分別處理不同的任務，Raspberry Pi 負責資料量較為龐大的影像處理部分，STM32F4-Discovery開發版執行與平板電腦的藍芽傳輸、光達的資料接收與馬達的控制，其中兩個微控制器以UART作為指令傳輸的方式。
本論文使用連續適應性均值飄移演算法(CAMShift)作為影像物件追蹤的方法，CAMShift擁有良好的效能與準確率，是一個廣泛使用的影像追蹤方法，透過連續的均值飄移演算法，並透過目標物件的特徵產生直方圖來計算最佳的物件位置，有鑑於過去的電動輪椅多為針對馬達控制演算法，較少針對影像辨識與安全而成的系統，故本論文透過影像辨識的結果，加上近年來由於自動駕駛而熱門的光達結合達到安全互動的目的。
經由使用者在APP上選擇的模式，可以分為自動追蹤模式與手動控制模式，自動追蹤模式可依據使用者所選的顏色追蹤具有該顏色的物件，經過Raspberry Pi上的影像處理結果，回傳給負責馬達控制的STM32F4-Discovery開發版，進而移動小車到該物並維持安全距離，可做為居家溝通互動的科技輔具之外，也能夠擴展到外出型的電動輪椅做小範圍的社區活動。
實驗結果顯示，由CAMShift計算之位置，透過兩控制器的資料傳輸，再經由魚眼相機的校正轉換從相機影像位置投影至實際的位置，可以成功鎖定目標物的位置，加上光達的安全距離輔助，可以達到穩定性與安全性。
本論文結合近年熱門的自動駕駛技術，以低功耗、低成本的方法設計科技輔具，以期望能夠增進重障者與家人、照護者的互動。

There are a lot of mature mobile assistive devices for increasing amount of elderlies and people with disabilities. But for highly disabled people, these mobile assistive devices have relatively high threshold for them. For those patients with the lack of ability, neither traditional joystick-controlling wheelchairs nor wheelchairs with sound control can’t offer a suitable environment for highly disabled people. As a result, this decrease the chance for outside activities and interactions with families for them, so caregivers are highly relied for them. So, a system combining image object tracking and the LiDAR obstacle avoidance is proposed. This system can be used in household environment to enhance the condition of interacting with families. Also, it is suitable for going outside with a smart electrical power wheelchair away from the home.
In this research, a system composed of embedded systems, an Android Tablet, a Light Detection and Ranging, and cameras with 180 degree fisheye lens is constructed. This system is suitable for image-tracking small tracked cars and can be migrated on the electrical power wheelchairs for outside activities. Using cameras with 180 degree fisheye lens can increase the field of view, and it’s helpful for people with high disabilities and low sensitivities. This project use a tablet PC which is available everywhere as the user interface, the Android APP on it display the whole required information while driving. For the division of work, two embedded system microcontrollers are responsible for different tasks respectively. The Raspberry Pi is in charge of image processing with a large amount of data calculation. On the other hand, the STM32F4-Discovery discovery kit execute the data transmission to the tablet PC through Bluetooth, LiDAR data receiving, and the motor control. UART is used to communicate between two microcontrollers.
This research uses the Continuously Adaptive Mean-Shift as the method of image object tracking. CAMShift has good efficiency and precision, and it widely used in image tracking. By doing Mean-Shift continuously and adaptively, it calculates out the best position according to the target color feature and its histogram. The motor control algorithm have been aimed at in wheelchair researches over the past years, and there are few wheelchair systems in connection to image recognition. This research adds the LiDAR sensor which is popular in the field of self-driving car to achieve the purpose of interacting safely.
Via the mode selected by users on the Android APP, it can be divided in the object tracking mode and the manually control mode. The object tracking mode tracks the objects with a certain color according to the color users choose. The results returned from the Raspberry Pi are passed to the STM32F4-Discovery discovery kit and then move the small tracked car toward the target and maintain the safety distance.
The experimental results shows that the positions CAMShift calculated are projected to actual position from camera image position through the calibration transform of fisheye lens camera. It successfully targets the best position of objects with specific color. By the assistance of LiDAR, the distance of safety is considered to reach the stability and the driving safety.
This research combines the techniques of self-driving, and the integrated science assistive system consisting of low power and low-cost sensors and embedded systems is designed. This system can be used in household communication and also be expanded to electrical power wheelchair to do social activities. The enhanced interaction with families and caregivers is expected.

[1] C. M. Wu and C. H. Luo. (2002). Morse Code Recognition System With Fuzzy Algorithm for Disabled Persons. J. Med. Eng. & Tech., Vol. 26, Num. 5, pp. 202-207, 2002.
[2] C. H. Liang, C. M. Wu, S. W. Lin, and C. H. Luo. (2009). A Portable and Low-cost Assistive Computer Input Device for Quadriplegics. Technology and Disability, vol. 21, no. 3, pp. 67-78, 2009.
[3] C. H. Luo, P. F. Chen, and L. G. Young. Zigbee Morse Code Embedded System Design for Wireless Digital House Appliances. Proceedings of 2012 TREATS Annual Meeting, p6. June, 2012.
[4] C. H. Luo, P. F. Chen, and L. G. Young. Zigbee Morse Code Embedded System Design for Wireless Digital House Appliances. Proceedings of 2012 TREATS Annual Meeting, p6. June, 2012.
[5] Y. C. Wang, Interactive Assistive Electric Wheelchair with Embedded System, 2013.
[6] J. Y. Shiao, Y. C. Wang, C. H. Luo, P. C. Chen, S. C. Chen and M.D. Shieh. (2015). Design and Approach of Interactive Assistive Electrical Power Wheelchair for Quadriplegic. Symposium on Engineering, Medicine and Biology Applications, Kaohsiung, Taiwan, Jan. 30 – Feb.1 2015.
[7] J. Y. Shiao, Neural Network Control and Driver Assistance System of Interactive Electrical Power Wheelchair for People with Quadriplegia, 2014.
[8] Z. Liu, Yangzhou Chen, Zhenlong Li. Camshift-Based Real-Time Multiple Vehicle Tracking for Visual Traffic Surveillance. Computer Science and Information Engineering, 2009 WRI World Congress on (Volume: 5), 2009.
[9] Q. Liu, C. Cai, K. N. Ngan, and H. Li. CAMShift Based Real-time Multiple Face Multiple Face Match Tracking. Proceedings of 2007 International Symposium on Intelligent Signal Processing and Communication Systems Nov.28-Dec.1, Xiamen, China, 2007.
[10] X. Chen, X. Li, H. Wu, and T. Qiu. Real-time Object Tracking via CamShift-Based Robust Framework. 2012 IEEE International Conference on Information Science and Technology Wuhan, Hubei, China; March 23-25, 2012.
[11] Christian Brauer-Burchardt, Klaus Voss. A new algorithm to correct fish-eye- and strong wide-angle-lens-distortion from single images. Image Processing, 2001. Proceedings. 2001 International Conference on (Volume: 1), 2001.
[12] Juho Kannala, Sami S. Brandt. A Generic Camera Model and Calibration Method for Conventional, Wide-Angle, and Fish-Eye Lenses. IEEE Transactions on Pattern Analysis and Machine Intelligence (Volume: 28, Issue: 8), 2006.
[13] Saurabh Ladha, Deepan Kishore Kumar, Pavitra Bhalla, Aditya Jain, R.K. Mittal. Use of LIDAR for Obstacle Avoidance by an Autonomous Aerial Vehicle. Third Symposium on Indoor Flight Issues, August, Grand Forks, 2011.
[14] Leandro C. Fernandes, Maurício A. Dias, Fernando S. Osório, and Denis F. Wolf. A Driving Assistance System for Navigation in Urban Environments. Lecture Notes in Computer Science Volume 6433, pp 542-551, 2010.
[15] Keonyup Chu, Minchae Lee and Myoungho Sunwoo. Local Path Planning for Off-Road Autonomous Driving With Avoidance of Static Obstacles. Intelligent Transportation Systems, IEEE Transactions, Vol. 13, No. 4, December 2012.
[16] RoboPeak Team. RPLIDAR Low Cost 360 Degree 2D Laser Scanner (LIDAR) System Introduction and Datasheet. Rev.6, April 2014.