AGV positioning and mapping method, system and device and medium

Abstract

The invention discloses an AGV (Automatic Guided Vehicle) positioning and mapping method, system and device and a medium. The method comprises the following steps: acquiring image information in an environment; extracting ORB feature points of each frame of image, and performing feature point matching on two adjacent frames of images according to the extracted ORB feature points; obtaining a rotation matrix and a translation vector between two adjacent frames of images according to the pixel positions of the paired ORB feature points; on the basis of a feature point matching result, parameterizing ORB feature points and camera poses by adopting a parallax angle parameterization method; constructing a back-end optimization objective function according to the rotation matrix, the translation vector, the parameterized ORB feature points and the camera pose, and optimizing the camera pose and the ORB feature points according to the back-end optimization objective function; and establishing an environment map according to the optimized camera pose and ORB feature points. The camera pose and the feature points are parameterized by adopting a parallax angle parameterization method, the problem of non-convergence of a back-end optimization algorithm is avoided, and the method can be widely applied to the technical field of AGV control.

Description

technical field [0001] The present invention relates to the technical field of AGV control, in particular to an AGV positioning and mapping method, system, device and medium. Background technique [0002] The rapid development of computer and sensor technology has largely promoted the research on localization and mapping (SLAM) of AGV mobile platform, and the visual SLAM algorithm using camera as sensor is low in cost, light and easy to carry, and can obtain rich It is favored by researchers due to its many advantages such as scene information. At present, visual SLAM is mainly divided into the SLAM algorithm based on the feature point method and the SLAM algorithm based on the optical flow method. Among them, the SLAM algorithm based on the feature point method has relatively good comprehensive performance, such as the ORB-SLAM2 algorithm, which not only has a faster processing speed, And it has higher accuracy. [0003] In the visual SLAM algorithm based on feature point...

AI Technical Summary

Problems solved by technology

At present, most visual SLAM algorithms use the XYZ coordinate system to parameterize the feature points. Although this parameterization method can express the spatial position of the feature points very intuitively, for some feature points that are far away from the camera or when the camera direction and When the AGV motion direction is on the same plane, the back-end optimization algorithm will not converge or the number of optimization iterations will increase significantly, resulting in slower or even unstable algorithm speed

The inverse depth parameterization method uses the reciprocal of the distance to represent the distance between the feature point and the camera center. Although it can handle the feature points with a long distance, it still cannot resolve the after-effects caused when the camera direction and the AGV motion direction are on the same plane. The problem of non-convergence of terminal optimization algorithm

Images