Simultaneous localization and mapping method of monocular vision robot based on retracking strategy

Abstract

The invention discloses a simultaneous localization and map construction method of a monocular vision robot based on a retracking strategy. The method comprises the following steps: 1) extracting ORB characteristics of each picture; 2) tracking camera postures by utilizing characteristic matching of the adjacent pictures; 3) carrying out relocation and retracking strategy on pictures which are not tracked; 4) if the relocation is successfully executed, estimating postures of a current camera; stopping the retracking strategy and deleting a temporary variable generated by the retracking strategy; 5) if the retracking strategy is successfully executed, generating a new track; 6) judging the quantity of the generated tracks; if the quantity is more than a threshold value, eliminating an initial track; 7) carrying out closed-loop detection on each key frame and carrying out track fusion after the detection is successful; 8) when a localization system is finished, screening the tracks to obtain one track with the accurate posture. The simultaneous localization and map construction method disclosed by the invention has the advantages that the complete camera track can be located under the conditions of rapidness in movement, shielding, insufficient textures and illumination changes.

Description

technical field [0001] The invention belongs to the field of autonomous positioning and navigation of mobile robots based on a single camera, and in particular relates to a monocular vision synchronous positioning and map construction method based on a re-tracking strategy. Background technique [0002] With the development of computer technology and artificial intelligence, the research of robots has received more and more attention and investment, and intelligent autonomous mobile robots have become an important research direction and research hotspot in the field of robotics. Simultaneous localization and mapping (SLAM) problem refers to a mobile robot in an unknown environment, using its own sensors (such as lidar, odometer, camera, ultrasonic, etc.) to create a map consistent with the environment, and At the same time, the map is used to estimate the position and attitude of the robot. This problem is considered to be the key issue of whether the robot can truly realiz...

AI Technical Summary

Problems solved by technology

[0006] In view of the technical problems existing in the prior art, the purpose of the present invention is to provide a method for synchronous positioning and map construction of a monocular vision robot based on a re-tracking strategy; Part of the camera pose and the environment cannot be estimated and modeled. An innovative re-tracking strategy is proposed to improve the success rate of camera pose estimation and the completeness rate of map reconstruction.

Images