Method and device of generating high-precision semantic map, and computer storage medium

Abstract

The invention relates to technologies of vehicle navigation and automatic driving, and particularly relates to a method and a device of generating a high-precision semantic map, a computer storage medium and a navigation map surveying and mapping vehicle comprising the device. The method of generating the high-precision semantic map comprises following steps of: acquiring a point cloud map relatedto surrounding environment and multiple images; determining a semantic recognition target according to a target recognized in the image and a corresponding target in the point cloud map, wherein thetarget is correlated with an object in a plane intersecting with the ground; extracting a static environment semantic characteristic from the point cloud map; storing the semantic recognition target and a descriptor relating to the static environment semantic characteristic in a positioning layer of the semantic map; generating, based on the image and the point cloud map, a precise geographic category probability map for extraction of ground graticule; and generating a high-precision semantic map planning layer by using the semantic recognition target and the precise geographic category probability map, wherein the planning layer comprises a road topological relationship, a lane topological relationship and a basic road indication information.

Description

technical field [0001] The invention relates to vehicle navigation technology and automatic driving technology, in particular to a method, device, computer storage medium for generating a high-definition semantic map, and a navigation map surveying vehicle including the device. Background technique [0002] Traditional navigation maps cannot meet the needs of autonomous driving due to insufficient accuracy. As a necessary part of unmanned driving, high-definition maps have gradually become the consensus in the industry, and they have the advantages of high precision and multiple dimensions. The high-definition map can provide the driving system with more forward-looking information indication and information redundancy, realize the matching and positioning of the car, enable the driving system to perceive a wider range of traffic situations, and ensure the safety of automatic driving. [0003] Among high-definition maps, point cloud maps are favored by the autonomous drivin...

AI Technical Summary

Problems solved by technology

However, the data volume of point cloud maps is huge, which brings great obstacles to map storage and online matching and positioning.

In recent years, although the industry has done research on sensor fusion and the internal semantic structure of point cloud maps, it has not yet provided a high-definition map generation method and device that can effectively reduce the amount of data while maintaining high precision.

Images