手眼協調(hand-eye coordination)能力為人類所具有的一項高等生物行為動作。而在人型機器人系統上，為達到模仿人類行為動作，手眼協調功能之建立為一重要的研究課題，其常需利用視覺伺服為基礎來完成。本論文旨在研究以視覺伺服為基礎，以倒單擺系統為一平台，結合CMOS (Complementary Metal-Oxide Semicond-uctor)影像感測器來模擬一手眼協調控制系統。

　　整個系統主要分為機械結構、馬達致動器、影像處理模組、靜態隨機存取記憶體(Static Random Access Memory, SRAM)、影像感測器和平衡控制器部分。其中，影像處理模組部份主要是以Altera公司所生產的可程式邏輯閘陣列(Field Programmable Gate Array, FPGA)來完成；影像感測器則是採用國家半導體公司(National Semiconductor)所生產的CMOS影像感測器；而平衡控制器部分是以德州儀器(Texas Instruments, TI)生產的數位訊號處理器(Digital Signal Processor, DSP) TMS320F240為控制核心。

　　旋轉型倒單擺系統為一快速、不穩定之非線性系統，由兩個不同平面的連桿所組成，其控制目的為由直流馬達驅動旋轉臂帶動單擺作平衡控制。在本論文中，量測單擺偏擺角度則是以影像感測器來取代以往所用的光學式編碼器來作為位置感測元件。其方法首先，利用CMOS影像感測器來擷取單擺的偏擺位置。之後，將擷取到的影像資料存入至外掛的記憶體中，再由FPGA將記憶體內的影像資料讀出並做相關的影像演算法處理。最後，將由FPGA所計算出的單擺偏擺位置資訊送入由數位訊號處理器為核心所構成的LQR平衡控制器中，此平衡控制器則依據單擺所偏擺的角度來完成單擺的平衡。經過驗證與分析後，本研究證實在以視覺伺服為基礎下可完成旋轉型倒單擺的平衡控制。

　Hand-eye coordination is a dexterity ability of the advanced creatures. To imitate the action of the human in humanoid robotics systems, the development of the hand-eye coordination ability is an important task. The visual servo techniques are always exploited for achieving this goal. The objective of this thesis is to control an inverted pendulum system with a CMOS (Complementary Metal-Oxide Semiconductor) image sensor that imitates a hand-eye coordination control system.

　The system consists of a inverted pendulum mechanism, a motor actuator, an image processing module, an SRAM, an image sensor, and a balance controller. In this system, the image processing module is implemented by an FPGA device (EP1C12Q240C8, Altera). The CMOS image sensor from National Semiconductor is used for sensing the displacement of the pendulum and the balance controller is implemented by a digital signal processor (TMS320F240, TI).

　The rotary inverted pendulum system is a fast, unstable and nonlinear system. It is composed by two rigid rods which lie in two different planes. The control objective is to balance the pendulum through the motor. To do so the digital image data of the image sensor is loaded into an SRAM. After that, the image processing algorithm is carried out by an FPGA device. Finally, the position of the pendulum obtained by the FPGA will be sent to a DSP-based LQR balance controller. This controller achieves balance control of the pendulum according to the angular displacement of the pendulum. The system is developed and proven to work well through simulation and experiment.

[1] L. G. Shapiro and G. C. Stockman, Computer Vision, Prentice Hall,Upper Saddle River, NY, 2001.
[2] D. J. Dailey, F. W. Cathey, and S. Pumrin, “An algorithm to estimate mean traffic speed using uncalibrated cameras,” IEEE Transactions on Intelligent Transportation Systems, Vol. 1, No. 2, pp. 98-107, 2000.
[3] A. J. Lipton, H. Fujiyoshi, and R. S. Patil, “Moving target classification and tracking from real-time video,” IEEE Workshop on Applications of Computer Vision, pp. 8-14, 1998.
[4] D. S. Jang, G. Y. Kim, and H. I. Choi, “Model-based tracking of moving object,” Pattern Recognition, Vol. 30, No.6, pp. 999-1008, 1997.
[5] G. D. Hager, W. C. Chang, and A. S. Morse, “Robot feedback control based on stereo vision: Towards calibration-free hand-eye coordination,” Proceedings of IEEE International Conference on Roborics and Automation, Vol. 4, pp. 2850-2856, 1994.
[6] K. Sabe and H. Hashimoto, “Real time tracking using active eye sensing system,” Proceedings of the 1995 IEEE Intermational Conference on Industrial Electronics, Control and Instrumentation, Vol. 1, pp. 156-160, 1995.
[7] B. K. P. Horn and B. G. Schunck, “Determine optical flow,” Artifical Intelligence, Vol. 17, pp. 285-294, 1981.
[8] B. Lucas and T. Kanade, “An iterative image registration technique with an application to stereo vision,” Proceedings of the DARAP Image Understanding Workshop, pp. 121-130, 1981.
[9] L. Wenzel, N. Vazquez, and D. Nair, “Computer vision based inverted pendulum,” Proceedings of the 17th IEEE Instrumentation and Measurement Technology Conference, Vol. 3, pp. 1319 -1323, 2000.
[10] K. Rasit, C. Serap, and O. Cemil, “Moving object detedtion and target prediction in video image using computer vision,” Pattern Recognition, Vol. 31, No. 3, pp. 526-529, 1998.
[11] S. Jorg, J. Langwald, and M. Nickl, “FPGA based real-time visual servoing,” Proceedings of the 17th International Conference on Pattern Recognition, Vol. 1, pp. 749-752, 2004.
[12] 繆紹綱，「數位影像處理」，普林斯頓國際有限公司，民國九十二年。
[13] P. Anandan, Measuring visual motion from image sequences, COINS Dept. Univ. of Massachusetts, Tech. Rep. COINS- TR-87-21, 1987.
[14] L. Xu and D. C. Hogg, “Neural network in human motion tracking-An experimental study,” Image and Vision Computing, Vol. 15, No. 8, pp. 607-615, 1997.
[15] K. J. Astrom and B. Wittenmark, Computer-Controlled Systems-Theory and Design, Prentice Hall, Upper Saddle River, NJ, 1997.
[16] K. Ogata, Modern Control Engineering, Prentice Hall, Upper Saddle River, NJ, 2002.
[17] 范逸之、陳立元，「Visual Basic與RS-232串列通訊控制」，文魁資訊股份有限公司，民國八十九年。
[18] KODAK KAC-9630 CMOS Image Sensor Device Performance Specification, KODAK, 2004.
[19] The -Bus Specification Version 2.1, Philips Semiconductors, 2000.
[20] IC61LV2568 256K 8 High-Speed CMOS Static RAM, ISSI Inc., 2000.
[21] SN74ALVC164245 16-bit 2.5V to 3.3V/3.3V to 5V Level-Shifting Transceiver With 3-State Outputs, Texas Instruments Inc., 2004.
[22] Cyclone Device Handbook, ALTERA Inc., 2002.
[23] 龔應時、陳建武、徐永松，「TMS320F/C24x DSP 控制器原理與應用」，滄海，民國九十年。
[24] 詹富強，「以數位信號處理器為基礎單板獨立旋轉型倒單擺甩上與平衡控制系統之實現」，國立成功大學工程科學系碩士論文，民國九十三年。
[25] P. I. Corke and M. C. Good, “Dynamic effects in visual-loop systems,” IEEE Transactions on Robotics and Automation, Vol. 12, No. 5, pp. 671-683, 1996.
[26] S. B. Park, “A motion detection system based on a CMOS photo sensor array,” Proceedings of IEEE Conference on Image Processing, pp. 967-971, 1998.
[27] D. Murray and A. Basu, “Motion tracking with an active camera,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 16, pp.449-459, 1994.
[28] 吳駖，「MATLAB 6.x 與基礎自動控制」，松崗，民國九十一年。
[29] 林卓毅，「具動作偵測功能之即時影像處理嵌入式系統」，國立台灣大學電機工程學系碩士論文，民國九十二年。
[30] 李建緯，「平面移動物體追蹤系統之研製」，國立交通大學電機與控制工程學系碩士論文，民國八十八年。