影像式單攝影機之機器人動態避障路徑系統
No Thumbnail Available
Date
2013
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
本論文提出一種影像辨識的方式，應用於自走車的障礙物偵測與其路徑規劃。首先透過二值化、中值濾波、形態學處理、canny運算、連通元件等影像處理方法，找出障礙物的位置，再利用所提出的平行線距離量測法，偵測影像中障礙物的實際座標。於實驗環境中平行線距離量測系統的架設，只需使用單一網路攝影機及四個參考目標點。因此實驗環境的架設建構相對的較簡單，而且架設成本相對的便宜。最後，將影像辨識方法與平行線距離量測結合，完成自走車的避障路徑規劃系統。在路徑規劃的方面，分為靜態的路徑規劃以及動態的路徑規劃，靜態的路徑規劃是將網路攝影機架設在天花板上，並將規劃的路徑即時的回傳於二維地圖上，做為自走車移動的依據；而動態的路徑規劃則是將網路攝影機架設在機器人上，使機器人在遇到障礙物時，能夠即時的進行避障路徑的規劃，以利於機器人的行走，最後以實驗驗證此方法的可行性。
This thesis proposes a way to identify the image which is applied to the obstacle-detecting and path-planning of a wheeled mobile robot. First of all, we use some image processing methods including medium filter, morphologic processing, Canny edge detector, and connected component labeling to find the locations of obstacles. Then we use a parallel lines distance measuring method to detect the actual coordinates of the obstacles in the image. To set up the parallel lines distance measuring system in the experiment, we just need a single web camera and four juxtaposed points. Thus the setting of this experiment is relatively easy and the cost is lower. Finally, we combine the image identifying method with parallel lines distance measuring method to complete the path-planning of a wheeled mobile robot system. To design the path-planning, we have two methods including static and dynamic states. The static path-planning method is to set up a web camera on the ceiling and instantly transfer the images to a two-dimensional map as a basis of the wheeled mobile robot’s movements. The dynamic path-planning method is to set up the web camera on the wheeled mobile robot and make it plan the path which can avoid the obstacles simultaneously. By the experiments, the wheeled mobile robot can go smoothly and we can illustrate the feasibility of the proposed method.
This thesis proposes a way to identify the image which is applied to the obstacle-detecting and path-planning of a wheeled mobile robot. First of all, we use some image processing methods including medium filter, morphologic processing, Canny edge detector, and connected component labeling to find the locations of obstacles. Then we use a parallel lines distance measuring method to detect the actual coordinates of the obstacles in the image. To set up the parallel lines distance measuring system in the experiment, we just need a single web camera and four juxtaposed points. Thus the setting of this experiment is relatively easy and the cost is lower. Finally, we combine the image identifying method with parallel lines distance measuring method to complete the path-planning of a wheeled mobile robot system. To design the path-planning, we have two methods including static and dynamic states. The static path-planning method is to set up a web camera on the ceiling and instantly transfer the images to a two-dimensional map as a basis of the wheeled mobile robot’s movements. The dynamic path-planning method is to set up the web camera on the wheeled mobile robot and make it plan the path which can avoid the obstacles simultaneously. By the experiments, the wheeled mobile robot can go smoothly and we can illustrate the feasibility of the proposed method.
Description
Keywords
障礙物檢測, 影像量測, PLDM, obstacle detecting, image measuring, PLDM