本論文提出一全方位距離測量與移動機器人避障演算法。實驗的場景以符合FIRA (Federation of International Robot-soccer Association) RoboSot規則之避障挑戰賽中的場地為主。全方位攝影機相較於傳統攝影機優點在於擁有寬廣的視野，能夠減少視覺死角與提高追蹤物體的能力，但是其視覺影像會有扭曲變形。因此本文提出一全方位影像測量方法，可以測量出雙曲面鏡之曲率與變焦鏡頭之焦距，將扭曲影像還原成平面影像。最後再利用全方位視覺影像的座標點與雙曲面鏡離地面之高度，即可測量出障礙物與機器人之間的距離。在避障挑戰賽中，足球機器人需避開障礙物往目標位置前進。本文根據動態視窗法與人工勢場法，提出一適合全方位移動平台的動態視窗演算法。從實驗結果中，可以得知所提的方法能有效且快速達到避障的效果。 This thesis proposes an omnidirectional distance measurement and an obstacle avoidance algorithm of the mobile robot. The experimental scenario was based on the venue of the obstacle avoidance challenge in accordance with the RoboSot rules of the Federation of International Robot-soccer Association (FIRA). Omnidirectional cameras have the advantage of a wide angle image, but the image will be distorted. It compares with the conventional cameras, which can reduce visual dead zones and improve the ability to track objects. Therefore, this thesis proposes an omnidirectional distance measurement method, which can measure the curvature of the hyperbolic mirror and the focal length of the zoom lens, and the project the distorted image to a flat image. Finally, the distance between the obstacle and the robot can be measured by using the coordinate point of the HyperOmni Vision image and the height of the hyperboloidal mirror from the ground. In the obstacle avoidance challenge, the soccer robot needs to avoid the obstacle and advance to the goal position. Therefore, this thesis proposes a dynamic window algorithm for the omnidirectional mobile robot based on the Dynamic Window Approach (DWA) and Artificial Potential Field (APF). From the experimental results, the algorithm can effectively and quickly achieve the effect of obstacle avoidance.
