This chapter delves into the intricacies of robot navigation and obstacle avoidance, focusing on visual relocalization and the challenges it presents. The chapter explores various classic relocalization algorithms, such as geometric methods and image retrieval, and their evolution with the incorporation of deep learning techniques. It also discusses the significance of camera pose in navigation, the role of feature extraction and matching in relocalization, and the application of RANSAC for robust pose estimation. Furthermore, it touches on the practical aspects of robot navigation, including map-based planning and the adaptation to dynamic environments. The chapter equips readers with foundational knowledge and practical codes to grasp the nuances of visual relocalization and robot navigation.