Mobile robot V-SLAM method of three-stage point cloud registration method

Abstract

The invention provides a mobile robot V-SLAM method of a three-stage point cloud registration method. The V-SLAM method comprises the following steps: S1, acquiring RGB information and Depth information of the surrounding environment; 2, generate three-dimensional point cloud data; 3, extract that ORB image feature from the obtained RGB image, and match the feature of the point set elements by adopting FLANN; S4, screening the point pairs of the RGB map through the RANSAC sampling strategy so as to obtain the interior points to complete the preprocessing stage; S5, completing the initial registration stage of the point cloud by adopting the corresponding point double distance threshold method based on the rigid body transformation consistency; S6, a dynamic iterative angle factor ICP precision registration method is introduced to complete the precision registration phase when the initial posture is good; S7, the key frame selection mechanism of sliding window and random sampling is introduced in the back end, and the g2o algorithm is used to optimize the robot posture trajectory, so as to realize the reconstruction of three-dimensional point cloud environment The invention can improve the registration accuracy and registration efficiency of the point cloud map in the environmental three-dimensional reconstruction.

Description

technical field [0001] The invention belongs to the field of mobile robot V-SLAM, especially a method for point cloud map reconstruction. Background technique [0002] In recent years, with the rapid development of mobile robot intelligence and computer vision processing technology, mobile robots are widely used in industries such as home life, catering, virtual reality and autonomous navigation. Visual simultaneous localization and mapping (V-SLAM) refers to the reconstruction of the surrounding environment by the visual sensor carried by the mobile robot and the estimation of its own position during the traveling process of the mobile robot. With the low cost and high precision of visual sensors, more and more applications are used in robots, especially the emergence of Kinect depth cameras, which have the advantages of obtaining rich environmental information and fast update speed, and are well applied in mobile robots V- In the SLAM system. [0003] V-SLAM is generally...

AI Technical Summary

Problems solved by technology

[0004] The present invention aims to solve the problems of easy falling into local optimum and large matching error in the existing point cloud registration algorithm

Images