Citation

Hong I, Park M. Trans. Kor. Soc. Automot. Eng. 2023; 31(12): 979-985.

Copyright

(Copyright © 2023, Korean Society of Automotive Engineers)

DOI

10.7467/KSAE.2023.31.12.979

PMID

unavailable

Abstract

To ensure safety in driving autonomous vehicles, the use of information tools, such as installing sensors like Lidar on the side of the roads or infrastructure, is being studied. In the case of Lidar installed in an infrastructure system, it can be calculated based on the characteristic that Lidar always looks in the same direction to separate it from the background. A Lidar sensor uses a large amount of Point Cloud Data(PCD), and it entails a large amount of computation because it must survey a large space. Also, a Lidar sensor is used for a long time, so the background might change. Thus, the algorithm must be updated. In this paper, algorithms that generate data to remove the background are placed in parallel to reduce the amount of computation. It also proposes a method of computing the background with data before a certain frame to note the change in the background. An infrastructure lidar was then installed and verified. It was shown that the amount of noise generated during background removal was reduced to about 35.3%, compared to the previous result.

Language: ko