SLAM method and system based on binocular event camera

Abstract

The invention provides an SLAM method and system based on a binocular event camera, and the method comprises the steps of carrying out the motion compensation of inputted left and right event camera data through IMU assistance, and obtaining a corresponding reconstructed image; using the IMU for assisting in motion compensation, projecting coordinates of an event point into a reference coordinatesystem through a relative pose obtained through IMU integration, and replacing the depth of the event point with the median of the depth of an adjacent three-dimensional space point; respectively carrying out feature point detection and tracking on corresponding reconstructed images input by the left and right event cameras; performing triangulation calculation on the detected and tracked featurepoints to obtain three-dimensional coordinate points of the corresponding target and pose changes between the images, and calculating a camera pose by using a PnP method; and performing back-end BA optimization in combination with IMU pre-integration to obtain a camera motion track and scene mapping information. The technical scheme of the invention can cope with scenes with large illumination change and high-speed motion, and can be used for solving the problem that the robot is easy to fail when moving or the environment is too complex in the existing SLAM system.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to a technical solution for realizing SLAM with a binocular event camera. Background technique [0002] Over the past few decades, interest in robotic perception has increased due to the research and development of computer vision methods. This traditional camera is capable of capturing high information content of the camera's surrounding environment and has become the most popular in various applications because of its low-cost, ubiquitous features. [0003] Simultaneous localization and map construction (Simultaneous Localization, SLAM) is one of the most important milestones in the field of robot perception, and has achieved remarkable achievements in the past 30 years. The monocular event camera SLAM cannot truly restore the scale, and the monocular camera needs a certain amount of time to initialize the camera, otherwise it will get wrong trajectory and mapping resu...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the problem that the existing SLAM method based on monocular event camera cannot directly recover t

Images