Road passable area detection method based on three-dimensional laser radar

Abstract

The invention provides a road passable area detection method based on three-dimensional laser radar. The road passable area detection method comprises the steps of: obtaining surrounding information of a vehicle by adopting three-dimensional laser radar, and collecting point cloud data; eliminating the motion distortion of the point cloud by combining the odometer information of the vehicle; extracting laser point cloud interest points according to the point cloud data after motion distortion is eliminated, wherein extracting laser point cloud interest points mainly comprises rejecting data points located out of an area at a certain height above the laser radar; performing ground segmentation by adopting the height information of the extracted laser point cloud interest points and combining the RANSAC algorithm, and distinguishing the ground point cloud and the obstacle point cloud; and rasterizing the distinguished obstacle point cloud, extracting data points closest to the vehicle in each grid, and combining the data points to obtain boundary points of the passable area. According to the invention, the problem of motion distortion generated during the motion of the point cloud is eliminated, so that the surrounding information expressed by the point cloud is more accurate.

Description

technical field [0001] The invention belongs to unmanned driving environment perception technology, in particular to a method for detecting a passable area of ​​a road based on a three-dimensional laser radar. Background technique [0002] Unmanned driving technology mainly includes environment perception, path planning and path tracking, among which environment perception is to use on-board sensors such as millimeter wave radar, camera and laser radar to perceive the surrounding information of the vehicle, so as to prepare for path planning and path tracking. In the sensor, the millimeter-wave radar cannot recognize the details of the object, and the recognition effect of the camera will be significantly affected in dark or backlight conditions, while the lidar can measure the details of the object and is not affected by the ambient light, so the lidar is in the environment perception It is widely used in the field and has a complementary effect with other sensors. [0003...

AI Technical Summary

Problems solved by technology

[0004] In the method based on the original point cloud, the invention patent with the patent number CN201810169285.X "A method for extracting the feasible region of structured roads based on 3D laser radar", and the invention patent with the patent number CN201810027774.1 "A method based on 3D laser The real-time road segmentation method of radar" is to extract the single-line point cloud from the multi-line lidar, and take the single-line point cloud as the research object. The former uses K-means clustering to optimize the extraction of obstacles, and combines the DBSCAN clustering algorithm to extract the roadside Obtain the feasible region, the latter uses the improved fuzzy line segment to generate the line segment map, and recognizes the road surface and obstacles to obtain the passable area. This kind of road is suitable for structured roads, and it is not universal in other road conditions such as the wild

Images