輪型機器人之即時影像追蹤系統控制與應用

Toward the Design and Control of Real-Time Image Tracking of a Wheel-Robot

邱鎮宇(Jeng-Yu Chiou)

指導教授 ： 陳建祥

國立清華大學/工學院/動力機械工程學系/碩士(2008年)

摘要

現今智慧型機器人的發展越來越趨於成熟。機器人可以藉由各類型的感測裝置，來判斷其身處的環境狀況，以進行下一步動作。影像感測器因具有低成本及短時間內取得高資訊量這兩大優點，故影像感測器已被廣泛的運用在智慧型機器人上。智慧型機器人應用的範圍也非常廣泛，包含家庭服務、休閒娛樂、醫療照護、導覽服務等。
本文以一個FPGA即時影像追蹤模組以及雙輪機器人為主要平台，搭配即時影像處理，建構出一個即時目標色球追蹤以及色球夾取的系統。實驗中利用CMOS攝影機模組擷取色球影像，擷取的影像經過影像色彩分類、色彩選擇、二維濾波以及影像二元化處理，接著便可根據處理過後的二元化影像得知目標色球的位置，然後根據影像所得到的目標色球位置資訊，來驅動雙輪機器人進行目標色球的追蹤。
正確追蹤到目標色球之後，利用建置在雙輪機器人上的機械夾爪，夾取所追蹤的目標色球，並將目標色球放置到指定的箱子內，即完成追蹤、夾取以及置放色球的任務。最後，將四種顏色色球依序追蹤以及抓取並且置放於相對應顏色的箱子裡。

關鍵字

雙輪機器人 ； 機械夾爪 ； 色球追蹤 ； 影像處理

並列摘要

Nowadays, the development of intelligent robot becomes mature. The robot can judge the surrounding and then act based on the sensors mounted on it. Having the advantages of low cost and high data rate in a short time, the image sensor has been applied in intelligent robot extensively. The applications of intelligent robot are widely, including household services, entertainment, health care and guided service.
Processing on a FPGA-based real-time image tracking module and wheel-robot, this thesis develops the method of a real-time colored-ball tracking and gripping system based on the real-time image. In the experimental setup, the CMOS camera module acquires the image of the targeted colored-ball. Through classification of color, the image selection of color, two-dimension filtering and the image binarization process, we are able to recognize the position of the target colored-ball by the binary image. Knowing the position data of colored-ball by the real-time image, the wheel-robot will start to track the target colored-ball.
After accurately tracking the target colored-ball, a mechanical gripper mounted on the wheeled-robot would begin to grasp the ball and put it in the corresponding box. Therefore, the tracking, grasping and releasing tasks are completed. Finally, tracking and grasping four colored-balls in order and releasing the each colored-ball to the matching box has been demonstrated through experiments.

並列關鍵字

wheeled-robot ； mechanical gripper ； tracking colored-ball ； image process

參考文獻

[16]俊源科技股份有限公司網站, http://www.juniortek.com.tw/#場景_1

[1]K. P. Horn and B. G. Schunck, ”Determining optical flow,”Artificial Intelligence, Vol.17, pp185-203, 1981.

[4]K. Nickels and S. Hutchinson, “Model-Based Tracking of Complex Articulated Objects,” IEEE Transactions on Robotics and Automation, pp. 28-36, February 2001.

[5]M. Kass, A. Witkin , and D. Terzopoulos, “Snake: Active Contour Models,” International Journal of Computer Vision, pp. 321-331, 1988.

[7]Ying Wu and Thomas S. Huang, “Color tracking by transductive learning,” IEEE Conference on Computer Vision and Pattern Recognition, pp.133-138, June 2000.