Visual navigation method for moving robot

Abstract

The invention discloses a visual navigation method for a moving robot. The method comprises the following steps: acquiring images of left and right cameras, and calibrating the cameras; processing thevideo to obtain a frame image, and preprocessing the obtained image by adopting an image enhancement method based on active illumination; performing target identification on the image based on a YOLOv3 network target detection method; performing stereo matching on the image according to an SGBM algorithm optimized by a least square fitting interpolation method based on the obtained bounding box where the target is located; and obtaining the category of the image and the position information of the image. The target object can be detected, the boundary pre-selection box where the target is located can be framed out, and a foundation is laid for subsequent stereo matching to obtain depth information and avoid obstacles. According to the method, only matching operation is carried out on thepre-selected frame image, so that the size of the three-dimensional matching image is reduced, the calculation efficiency of the three-dimensional matching is greatly improved, and a beneficial effectis achieved for improving the real-time performance of target detection.

Description

technical field [0001] The invention relates to a visual navigation method for a moving robot, which belongs to the technical field of robots. Background technique [0002] Factories often use robots to carry workpieces and goods. In order to further realize the intelligent movement of the handling robot, it is necessary to use various sensing devices to enable the handling robot to recognize and judge the current road conditions and motion status, and make self-decisions, so that it can Make the handling robot realize completely unmanned operation, and run according to the prescribed route during the work process, or according to the task target, after identifying the environment, plan a reasonable movement trajectory by itself, and complete the intelligent operation. With the development of sports robots, their application fields are becoming more and more extensive. At present, the navigation methods of sports robots mainly include magnetic stripe navigation, inertial nav...

AI Technical Summary

Problems solved by technology

Magnetic strip navigation needs to lay magnetic strips on the moving path of the robot, the accuracy is low, and the magnetic strip protruding from the ground is easily damaged; with the accumulation of time, the cumulative error of inertial navigation increases, and other equipment needs to be assisted to correct it, and the high The cost of high-precision inertial navigation devices is relatively high; laser navigation needs to add reflectors on both sides of the moving path, which requires high installation accuracy of the reflectors, and is sensitive to other light sources, which is not easy to operate outdoors, and the cost is high

Images