Milemeter method based on fusion of laser SLAM and monocular SLAM

Abstract

The invention provides a speedometer method based on fusion of laser SLAM (Simultaneous Localization and Mapping) and monocular vision SLAM. Firstly, a laser radar and a monocular camera are adopted to collect data at the same time, then feature point extraction and matching are carried out on an image collected by the monocular camera, then point cloud of the laser radar is projected on the image collected by the monocular camera, and therefore the depth of feature points extracted before is calculated. When the priori pose is estimated, the depth information of the feature points in the previous frame is judged, and the pose is estimated through three different conditions, so that the more accurate priori pose is obtained, and finally, the final optimized pose is obtained through Bundle AdJustlement. The problems that in monocular SLAM, when a robot moves in the optical axis direction, pose estimation is inaccurate, and laser SLAM fails in the environment with similar geometric features are greatly solved.

Description

technical field [0001] The invention belongs to an odometer fusion method in SLAM, in particular to an odometer method based on the fusion of laser SLAM and monocular vision SLAM. Background technique [0002] Simultaneous localization and mapping (SLAM) is the basis for mobile robots to solve various problems such as exploration, detection, and navigation in unknown environments. According to the different sensors installed, the mainstream SLAM methods are divided into laser SLAM and visual SLAM. [0003] For visual SLAM, the camera has the advantages of low cost and rich image information, so it performs well in environments with similar geometric structures and processing loop detection and other issues. However, it is easily affected by illumination changes, especially when a monocular camera is used as a sensor, problems such as scale drift will occur. [0004] Laser SLAM is more mature in theory than visual SLAM, but it has problems such as poor effect in environment...

AI Technical Summary

Problems solved by technology

However, it is easily affected by illumination changes, especially when a monocular camera is used as a sensor, problems such as scale drift will occur.

[0004] Laser SLAM, the theory is more mature than visual SLAM, but there are problems such as poor effect in environments with similar geometric structures and difficulty in handling dynamic objects

Images