Design and evaluation of visual SLAM method based on Realsense for mobile robots

In recent years, with the popularization and rapid development of various types of robots in production and life, robots have become one of the most promising areas of research in the 21st century. The study of autonomous mobile robots in unknown environments is very challenging, with simultaneous localization and mapping (SLAM) holding the key to the development of truly autonomous mobility. The acquisition of depth images containing distance information about local objects is a vital component of SLAM, so the use of depth cameras as a new type of sensor is increasingly common. In this study, the RealSense R200 depth camera was used as an external environment sensor that enables the simultaneous positioning of robots in an unknown environment and map construction. A visual SLAM method was designed and verified through a series of experiments. The results demonstrated that the constructed map reflects obstacle occupancy information with high accuracy.