A Comparison of Two Approaches for Collision Avoidance of an Automated Guided Vehicle Using Monocular Vision

In this paper a comparison of two approaches for collision avoidance of an automated guided vehicle (AGV) using monocular vision is presented. The first approach is by floor sampling. The floor where the AGV operates, is usually monotone. Thus, by sampling the floor, the information can be used to search similar pixels and establish the floor plane in its vision. Therefore any other objects are considered as obstacles and should be avoided. The second approach employs the Canny edge detection method. The Canny edge detection method allows accurate detection, close to real object, and minimum false detection by image noise. Using this method, every edge detected is considered to be part of an obstacle. This approach tries to avoid the nearest obstacle to its vision. Experiments are conducted in a control environment. The monocular camera is mounted on an ERP-42 Unmanned Solution robot platform and is the sole sensor providing information for the robot about its environmen