Cited 0 time in webofscience Cited 0 time in scopus

Robust road curb detection using LiDAR and its application in real time vector map generation

Title
Robust road curb detection using LiDAR and its application in real time vector map generation
Authors
Kumar, Gurumadaiah AjayLee, Jin-HeeHwang, JongrakPark, Sung MinKim, DongukKwon, Soon
DGIST Authors
Lee, Jin-Hee; Kwon, Soon
Issue Date
2019-11-16
Citation
2019 대한임베디드공학회 추계학술대회, 328-331
Type
Conference
ISBN
9788996655312
Abstract
Acurate road curb extraction from the unorganized, noisy and huge point cloud data is chalenging task in the autonomous driving technology. Road curb features very useful in path planing and localization for the autonomous vehicle. Another important task that directly benefits from the curb detection is road feature map generation (vector map), which is one of the major requirement for the autonomous driving system. This paper presents an acurate, robust and eficient method to extract road curb points from the unorganized point cloud and generate the vector map by extracting the feature with in the road boundary. Road curb points are detected in two consecutive steps: First, based on the radius of each ring, points are filtered and filtered points are observed to extract smoth elevated points using a sliding window approach. Next, estimated road curb points are pased through the regresion filter to eliminate the detected non-curb points. In addition to curb points detection, paper presents prototype approach to generate the vector map based on the features with the road boundary.
URI
http://hdl.handle.net/20.500.11750/14148
Publisher
대한임베디드공학회
Related Researcher
  • Author Kwon, Soon  
  • Research Interests computer vision; deep learning; autonomous driving; parallel processing; vision system on chip
Files:
There are no files associated with this item.
Collection:
Division of Automotive Technology2. Conference Papers


qrcode mendeley

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

BROWSE