- 學位論文
以光流導引為基礎之機器人的設計與實現
Design and Implementation of Optical-Flow-Based Guiding Robot
摘要
本論文提出並實現一種以光流演算法為基礎之機器人視覺追蹤系統。透過使光流演算法來偵測場景中的變化，讓機器人可以隨場景中的揮手動作來做移動，達到視覺跟隨之目的。本論文將系統架構於Linux系統上，並以OpenCV函式庫來建立光流之擾動偵測演算法（optical-flow perturbation detect algorithm）。光流擾動偵測演算法主要有四步驟：（1）取得影像之差異，（2）去除背景雜訊，（3）計算畫面內之擾點能量，及（4）取得影像感興趣區域。在機器人的設計與實現上，本論文設計了一台可以穩定移動之機器人平台，讓光流演算法可以順利的在場景中偵測到揮手的動作變化，並且以硬體描述語言將比例積分控制器設計實現在FPGA晶片上來控制馬達。從實驗結果可知，本論文所提方法確實可以讓機器人有效的依據光流演算法來偵測揮手動作，達到以光流導引機器人的目的。
並列摘要
In this thesis, an optical-flow based tracking system is implemented. With the optical-flow perturbation detect algorithm, the robot can follow the hand waving in scene, and reach the goal of visual following. The system architecture in this thesis is based on linux system, and use OpenCV library to implement optical-flow perturbation detect algorithm. There are 4 steps of optical-flow perturbation detection: (1) Get difference between two frames, (2) Remove background noise, (3) Calculate the sum of perturbation energy, (4) Gather the Region Of Interest (ROI) from image. For design and implement action, this thesis designs a robot to move in constant speed, improve the result of optical-flow algorithm to detect the waving hand, and implement a Proportional and Integral Controller by Verilog HDL on FPGA to control the motor. From the experiment result, the method of this thesis can lead the robot to follow the waving hand, reach the goal of optical-flow tracking.