Indoor environment 3D semantic map construction method based on point cloud deep learning

Abstract

The invention relates to an indoor environment 3D semantic map construction method based on point cloud deep learning. The indoor environment 3D semantic map construction method comprises the following four parts: (1) acquiring a color map and a depth map of an indoor environment by using a depth camera; (2) constructing a point cloud deep learning network to obtain 3D semantic information of an object in the environment; (3) detecting a dynamic object and removing dynamic feature points; (4) solving a camera motion realization visual odometer, and constructing and optimizing a local map; (5)constructing a target semantic library through the obtained 3D point cloud semantic information; and (6) performing semantic fusion on the local map according to the target semantic library, and constructing an octree semantic map. Compared with a conventional method, the method has the advantages that the feature points in the dynamic object mask are eliminated by combining the semantic categoryinformation, the influence of the dynamic object on positioning and mapping is effectively reduced, the adopted 3D semantic acquisition mode is more direct and efficient, and better positioning and mapping capacity and semantic perception effect are achieved.

Description

technical field [0001] The invention belongs to the application of computer vision technology in the field of mobile robots, and specifically relates to a method for building a 3D semantic map of an indoor environment based on point cloud deep learning. Background technique [0002] Simultaneous Localization and Mapping (SLAM) algorithm is the process of simultaneously realizing the sensor's own pose estimation and environment map construction based on sensor data information. key technologies. According to different types of sensors, SLAM is divided into two categories: laser-based SLAM and vision-based SLAM. Among them, the research on SLAM based on laser sensors is relatively early, and the theory and technology are relatively mature. However, high-precision laser sensors are expensive and cost-effective. . The SLAM scheme using cameras as sensors is called visual SLAM (Vision-based SLAM, VSLAM). Traditional visual SLAM focuses on geometric positioning and mapping, whi...

AI Technical Summary

Problems solved by technology

[0007] In order to solve the problems of existing visual SLAM map construction methods, such as poor adaptability to dynamic scenes, lack of semantic awareness or complex and indirect semantic acquisition methods, the constructed point cloud map occupies a huge storage space, and the non-topological map structure cannot be directly used for navigation. For problems such as obstacle avoidance, the present invention provides a method for building a 3D semantic map of an indoor environment based on point cloud deep learning, the method comprising the following steps:

Images