This document outlines advancements in visual odometry (VO) and simultaneous localization and mapping (SLAM) using deep learning techniques. Key contributions include the development of end-to-end frameworks like DeepVO and UndeepVO, which leverage deep neural networks for real-time pose estimation and depth recovery. Additional projects such as Vinet and DeepFusion focus on fusing visual and inertial information for improved navigation and dense 3D reconstruction capabilities.