Lane line detection method, data processing unit and vehicle-mounted laser radar system

Abstract

The invention provides a lane line detection method. According to the method, extraction and noise reduction are carried out on lane line data through the modes of point data screening, point data filtering, adjacent point data clustering, miscellaneous point data removal, distance clustering, removal of point data sets corresponding to sidewalks and arrows, and direction clustering. According to the preferable embodiment of the invention, the point cloud data of laser radar echoes is directly utilized, so that the problems of data missing, data conversion errors and the like in the data conversion process are effectively avoided, and the problems of large data error, low data accuracy and the like caused by indirectly acquiring lane line information by utilizing reference positions are effectively avoided; a lane line detection result is obtained through multi-layer filtering and layer-by-layer optimization; and finally a lane line is drawn through curve fitting. Thus, the lane line detection accuracy is improved, errors are reduced, the extraction speed is higher, and the range of application scenes is wider.

Description

technical field [0001] The invention relates to the technical field of laser detection, in particular to a method for detecting lane lines by using a laser radar, a data processing unit of the laser radar, and a vehicle-mounted laser radar system. Background technique [0002] With the development of unmanned driving technology, lane line detection, as an important task of the environment perception system in the unmanned driving process, has gradually become a research hotspot. Only by accurately obtaining the specific position of the lane line, the autonomous navigation system can plan a reasonable driving path, and the unmanned vehicle can correctly make driving behaviors such as driving, overtaking and changing lanes. At present, the lane line detection method based on vision detection and the lane line detection method based on lidar detection are the two main methods for detecting lane lines. Since the visual detection method is sensitive to ambient light, it has relat...

AI Technical Summary

Problems solved by technology

[0004] The methods in the prior art generally convert point cloud data into rasterized data or histogram image data, instead of directly using the point cloud data measured by lidar, there will be missing data and data conversion errors during the data conversion process and other issues; in addition, the point data belonging to the ground is used to obtain the position of the roadside point as a reference position, and the position of the lane line is calculated according to the position of the roadside point. This method of indirectly obtaining lane line information by using the reference position will have large data errors and accurate data. issues of low sex

Images