Keywords: 
• visual SLAM (simultaneous localization and mapping) /
• deep learning /
• localization /
• mapping /
• semantic SLAM

Abstract: Simultaneous localization and mapping (SLAM) technique is an important research direction in robotics. Although the traditional SLAM has reached a high level of real-time performance, major shortcomings still remain in the positioning accuracy and robustness. The geometric environment map constructed by the traditional SLAM can satisfy the pose estimation of the robot. However, the same map is not good enough to support the robot to complete the self-navigation and obstacle avoidance. The interactive performance of such map is insufficient. With the development of deep learning technology, combining deep learning methods and SLAM to add semantic information has become an important research direction for the practical application of SLAM. In this paper, the latest researches about semantic information applied to SLAM are summarized. In addition, the prominent achievements of the semantics combined with the traditional visual SLAM of localization and mapping are introduced. Furthermore, the semantic SLAM is compared with the traditional SLAM in detail. Finally, the future research directions of advanced semantic SLAM are discussed.