A visual slam method based on semantic optical flow and inverse depth filtering

Abstract

The present invention relates to a visual SLAM method based on semantic optical flow and inverse depth filtering, comprising the following steps: (1) a visual sensor collects images, and performs feature extraction and semantic segmentation on the collected images to obtain extracted feature points and semantic Split results. (2) According to the feature points and segmentation results, the semantic optical flow method is used to initialize the map, and the dynamic feature points are eliminated to create a reliable initialization map. (3) Using the inverse depth filter on the initialized map to evaluate whether the 3D map points in the map are dynamic points, and expand the map according to the evaluation result of the inverse depth filter. (4) Continue to track, local map and loop detection in sequence for the extended map of the depth filter, and finally realize the visual SLAM for dynamic scenes based on semantic optical flow and inverse depth filtering.

Description

technical field [0001] The present invention relates to a visual SLAM method based on semantic optical flow and inverse depth filtering, which is a new visual SLAM method that combines semantic optical flow and inverse depth filtering technology, and is suitable for solving the problem of traditional visual SLAM systems in high dynamic scenes Failure and lack of understanding of the scene. Background technique [0002] Simultaneous Localization and Mapping (SLAM) refers to the estimation of the pose of the robot itself through the acquired sensor data without the prior information of the environment, and at the same time constructing a globally consistent environment map. Among them, the SLAM system based on visual sensors is called visual SLAM. Because of its low hardware cost, high positioning accuracy, and the advantages of completely autonomous positioning and navigation, this technology has attracted wide attention in the fields of artificial intelligence and virtual re...

AI Technical Summary

Problems solved by technology

[0005] The problem solved by the technology of the present invention is: to overcome the deficiencies of the existing technology, aiming at the problem that the visual positioning system of the system is susceptible to interference under dynamic scene conditions, a visual SLAM method based on semantic optical flow and inverse depth filtering is provided to improve the SLAM system to cope with dynamic scenes ability, improve the system's ability to understand the scene, and improve the system's positioning accuracy in dynamic scenes

Images