Method for completing visual SLAM closed-loop detection by fusing semantic information

Abstract

The invention discloses a method for completing visual SLAM closed-loop detection by fusing semantic information, and the method comprises an object recognition step of recognizing an object in an image through a target detector, and a stereo matching step of taking the recognized object as a stereo matching element according to related constraint conditions, a three-dimensional position estimation step of completing the estimation of the three-dimensional position of the object by using the mathematical model of the binocular camera; an attribute graph model construction step of extracting the object in the image and the spatial distribution information of the object and establishing a corresponding attribute graph set for the image set; and a similarity measurement step of matching and comparing the two attribute graphs. According to the method, the target detector is used for detecting and identifying the object in the image, while the existing image-oriented object detection methodhas good real-time performance, the subsequent calculation step is completed based on the identification result, the amount of data participating in the operation is very small, and the correspondingcalculation amount is relatively small, so that the method has good real-time performance, and good robustness can be maintained under a complex illumination condition.

Description

technical field [0001] The invention belongs to the field of map creation, and in particular relates to a visual SLAM closed-loop detection method with fusion of semantic information. Background technique [0002] In simultaneous localization and map creation (SLAM), closed-loop detection refers to judging whether the robot is in a certain area visited before according to the information obtained by the sensor, or whether the current location of the robot is in the created map. With a corresponding description. In the SLAM method based on graph optimization, closed-loop detection is a very critical link. Correct closed-loop detection helps to correct the odometer error, so as to obtain a map with small error and consistent global information, but wrong closed-loop detection will increase Errors can even destroy the entire map. [0003] The current mainstream loop closure detection method is based on key frames. One of the very common image matching methods based on visual...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to solve the problem of cumulative error of the visual sensor that occurs during the simultaneous positioning of the robot and the process of map construction, and to provide a closed-loop detection problem for the robot under complex lighting conditions, reducing the uncertainty caused by the robot pose and map variables Visual SLAM closed-loop detection method based on the fusion of semantic information for errors caused by sex

Images