Abstract
Mobile robots are used in a variety of fields, such as scientific research, industrial manufacturing, household services, and more. As robots tackle more complex tasks, the ability of robots to localize and map becomes more testable. Compared with GPS positioning, SLAM technology has more advantages in indoor and other environments. This article will combine the book "Probabilistic Robot" to introduce several traditional SLAM algorithms: EKF-SLAM, GRAPH-SLAM, FASTSLAM, ORB-SLAM… and propose some research on semantic SLAM. This paper mainly shows the difference and advantages between semantic SLAM and other SLAM.