Survey of Visual SLAM Based on Deep Learning

The review covers the latest research results of deep learning techniques applied to the field of SLAM(simultaneous localization and mapping), focusing on and summarizing the research results of deep learning in front-end tracking, back-end optimization, semantic mapping and uncertainty estimation, and looking forward to the development trends of visual SLAM under deep learning. This work can help the successors to understand and apply the deep learning techniques to studying the feasible solutions to the problem of autonomous localization and mapping for mobile robots.