Unstructured Road Slope Recognition Based on Improved RANSAC Algorithm

Liang Hong, Lijin Han, Hui Liu

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Unstructured road scenes usually have bumpy and undulating features, and there are often obstacles such as potholes and stones on the ground. At the same time, the lidar point cloud with low beams is relatively sparse, which easily affects the accuracy of slope detection results. In response to these problems, this paper proposes a real-time detection method for unstructured road slope based on the improved RANSAC algorithm. Use inertial navigation to obtain vehicle motion information, fuse historical frame point clouds to enrich the point cloud density, then perform multi-scale rasterization on the point cloud, and then use the improved RANSAC algorithm to fit the point cloud, and finally get the slope of the road. Experimental results verify that the algorithm can improve detection accuracy, real-time performance, and effectiveness.

Original languageEnglish
Title of host publication2023 5th International Conference on Communications, Information System and Computer Engineering, CISCE 2023
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages322-325
Number of pages4
ISBN (Electronic)9798350326796
DOIs
Publication statusPublished - 2023
Event5th International Conference on Communications, Information System and Computer Engineering, CISCE 2023 - Guangzhou, China
Duration: 14 Apr 202316 Apr 2023

Publication series

Name2023 5th International Conference on Communications, Information System and Computer Engineering, CISCE 2023

Conference

Conference5th International Conference on Communications, Information System and Computer Engineering, CISCE 2023
Country/TerritoryChina
CityGuangzhou
Period14/04/2316/04/23

Keywords

  • 3D LiDAR
  • IMU
  • Slope Detection
  • Unstructured Road

Fingerprint

Dive into the research topics of 'Unstructured Road Slope Recognition Based on Improved RANSAC Algorithm'. Together they form a unique fingerprint.

Cite this