Simultaneous Localization and Mapping Based on Probabilistic State Estimation

With the development of visual navigation, people demand more accuracy about SLAM problems. In this paper, we introduce the relevant contents of SLAM problems, such as Bayesian filters and their implementation forms, and basic methods to solve SLAM problems. Through the comparison and analysis of various methods, it is found that VSLAM can improve the accuracy of SLAM by obtaining many information from the camera and finally achieve good results. Finally, this paper also less some cutting-edge applications of SLAM technology represented by VSLAM based on deep learning, and discusses some methods to improve the performance of VSLAM. This paper discusses and categorizes some methods to solve the SLAM problem and summarizes the methods to improve the performance of VSLAM, so as to lay the cornerstone for the later study of SLAM accuracy problem.