Topological Map Generation Method Based on Visual Fusion Landmarks

Abstract

The invention discloses a topological map generation method based on visual fusion landmarks, which includes: inputting an RGB image, performing semantic segmentation on the RGB image based on a convolutional neural network, and extracting points whose output value is greater than a predetermined value as features according to the feature map of the middle layer of the semantic segmentation Points; obtain the depth map of the scene, based on the depth information of the depth map and the two-dimensional coordinate information of the feature points on the image, and obtain the three-dimensional coordinates of the feature points and the pose information of the robot according to the camera model to establish a three-dimensional map of the scene. Image texture segmentation based on deep learning to obtain the encoding of each pixel point in the texture feature space; using a landmark-level data fusion method based on fuzzy mathematics to obtain the membership distribution function of the point cloud relative to the predetermined landmark, according to the semantic membership distribution function and 3D map to obtain a 3D semantic map of fused landmarks; construct a topological map based on the 3D semantic map of fused landmarks to generate a semantic topological map of fused landmarks.

Description

technical field [0001] The invention belongs to the technical field of robots, in particular to a method for generating topological maps based on visually fused landmarks. Background technique [0002] Mapping technology is crucial for intelligent robots to explore the task environment and complete various tasks autonomously. In the process of movement, intelligent robots collect environmental information through sensors, analyze data, and perceive the environment, which helps to make corresponding decisions autonomously according to the needs of the task and the current state of the environment, and then complete various tasks to achieve a true sense of reality. intelligent. In the field of robotics, SLAM (simultaneous positioning and composition) technology is a key technology for robots to build maps and perceive the environment. During the process of traveling, the robot uses sensor data such as lidar and camera to estimate the position and attitude at this time, and b...

AI Technical Summary

Problems solved by technology

[0006] Aiming at the problems existing in the prior art, the present invention proposes a topological map generation method based on visually fused landmarks, which can run on embedded airborne or vehicle-mounted platforms. The 3D map constructed in the field does not contain semantic information, and at the same time solves the problem that the pixel-level semantic information fused in the visual SLAM system is easily affected by lighting and other conditions, and cannot realize language-based human-computer interaction

Images