ABSTRACT
In recent years, navigation has been a hot topic in the field of robotics, and machine vision is the most common method of autonomous navigation. Vision-based robot navigation is first studied, based on the three corresponding areas of pattern recognition, pattern matching and 3D reconstruction. Then the current situation and prospects for various types of algorithms are discussed, which include the algorithms of obstacle detection, stereo matching and Simultaneous Localization And Mapping (SLAM). The advantages and disadvantages of different algorithms are summarized. Finally, the prospects for robot navigation technology are discussed, and the algorithm framework that may be useful for related research fields is summarized.
