Construction method of indoor 3D point cloud semantic map

Abstract

The invention relates to the technical field of map construction, and relates to a construction method of an indoor 3D point cloud semantic map. The construction method comprises the following steps:1, acquiring an RGB-D image of an indoor environment by utilizing an RGB-D camera; 2, constructing a deep convolutional neural network Mask R-CNN which can be used for target detection and instance segmentation; 3, inputting the acquired RGB-D image into a network, and then performing point cloud processing on an output image; 4, fusing the image processed by the Mask R-CNN network and the point cloud with the constructed point cloud map of the indoor environment to obtain a semantic point cloud map; and 5, carrying out global optimization on the point cloud semantic map.

Description

technical field [0001] The invention relates to the field of deep learning and the field of map construction technology, in particular to a method for constructing an indoor 3D point cloud semantic map. Background technique [0002] The semantic perception ability of robots is one of the core focuses of current research on mobile robots (such as service robots, rescue robots, etc.). Traditional point cloud maps play a pivotal role in mobile robot movement (including but not limited to: unmanned driving, human-computer interaction, robot navigation, etc.) and path planning. However, due to the lack of semantic information of objects in traditional point cloud maps, robots lack the ability to understand the environment. Therefore, in recent years, some scholars have proposed the construction of 3D point cloud semantic maps. The point cloud semantic map refers to the point cloud map that contains the semantic information of the environment, such as tables, displays, bottles, k...

AI Technical Summary

Problems solved by technology

However, due to the lack of semantic information of objects in traditional point cloud maps, robots lack the ability to understand the environment. Therefore, in recent years, some scholars have proposed the construction of 3D point cloud semantic maps.

Images