This thesis is concerned about the visual navigation methods that rely on teach and repeat -technique. In these navigations, the robot is first controlled manually to follow a desired path while recording a video to create a visual path. Then, the robot is placed to beginning of the path to follow it autonomously according. Quantitative and Qualitative methods approaches are suggested for following this visual path. Qualitative methods require fewer calculations. However, the current qualitative methods have some shortcoming. In this research, the limitations of the qualitative methods and in particular limitations of the funnel path method are investigated. Then, a new approach, called the slopped funnel method, is introduced which overcomes the shortcoming. Several different scenarios are performed on a real robot to show the performance of the proposed method. In order to improve these types of navigations, a method to create a visual path and a method for visual localization are also proposed. In continuum, a novel visual navigation method based on Braitenberg's approach to control the robot over the visual path is introduced. This method, despite its efficiency, does not have the complexity of other qualitative methods and experiments performed to demonstrate the effectiveness of this approach.
