Visual inertia odometer method based on point-line initialization

Abstract

The invention provides a visual inertial odometer method based on point-line initialization; the method comprises the following steps: step 1, collecting visual data through a camera, and collecting inertial data through an IMU; step 2, performing image feature extraction and tracking: extracting any two frames of images and tracking point features and line features to obtain a feature corresponding relationship between front and back frames; step 3, performing IMU pre-integration: carrying out pre-integration on all IMU data between any two frames of images to obtain an IMU prior value, and if IMU zero offset changes, recalculating an IMU pre-integration measurement value according to new zero offset; step 4, performing initialization: sending the features extracted and tracked in the step 2 and the IMU pre-integration value calculated in the step 3 into a visual inertia initialization model to obtain an initial parameter value; step 5, performing tight coupling visual inertial positioning: sending the initial parameters to a tight coupling visual inertial estimator to realize real-time positioning of the carrier; according to the method, the initialization success rate of the system in a texture-free environment is increased, and the initialization time is shortened.

Description

technical field [0001] The invention belongs to the field of SLAM (Simultaneous Localization and Mapping, synchronous positioning and mapping), and specifically relates to a visual inertial odometer method based on point-line initialization. Background technique [0002] Due to the low cost of cameras and IMU (Inertial Measurement Unit, Inertial Measurement Unit) sensors and the complementarity of information collection capabilities, VIO (Visual-Inertial Odometry, Visual-Inertial Odometry) has been widely used in areas ranging from robot navigation to autonomous driving. . As the initial step of VIO, the initialization process can provide key information such as the scale, the pose and velocity of the first few image frames, the IMU zero bias, and the direction of the acceleration of gravity for the backend. Therefore, a robust VIO system must require an accurate and fast-converging initialization part to provide the initial information required by the backend. [0003] VI...

AI Technical Summary

Problems solved by technology

[0004] However, in some special texture-free environments, the VIO system cannot extract enough feature points in the scene, that is, there is no effective scene structure aligned with the IMU pre-integration measurement value, which eventually leads to the failure of the initialization process and even the divergence of the entire VIO system.

Images