With the development of the Internet, new technologies such as big data, artificial intelligence, virtual reality, block chains and quantum communications are emerging regularly and progressing. Augmented reality, as one of the important frontier directions of the new generation of information science and technology, integrates multi-media, sensors, new display technology, the Internet and artificial intelligence, and is gradually becoming a new focus of innovative applications in various industries. This paper details the development of VSLAM technology over recent years, and its application in augmented reality. We describe VSLAM’s development status, the classical schemes, and the specific system architecture of each scheme. The schemes are distinguished by their respective approaches to the four basic algorithmic frameworks: visual odometry, optimization, loop closing and mapping. We summarize current developments and existing problems and also consider the prospects.