Loading…
Road Boundary Detection Using Multi-channel LiDAR Based on Disassemble-reassemble-merge Algorithm for Autonomous Driving
To ensure the safe operation of self-driving cars, it is necessary to identify the location and layout of the road around the vehicle. Moreover, because road accidents pose a considerable risk to human life, accurate environmental awareness is paramount in autonomous driving, especially under unfavo...
Saved in:
Published in: | International journal of control, automation, and systems automation, and systems, 2023-11, Vol.21 (11), p.3724-3733 |
---|---|
Main Authors: | , , |
Format: | Article |
Language: | English |
Subjects: | |
Citations: | Items that this one cites Items that cite this one |
Online Access: | Get full text |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | To ensure the safe operation of self-driving cars, it is necessary to identify the location and layout of the road around the vehicle. Moreover, because road accidents pose a considerable risk to human life, accurate environmental awareness is paramount in autonomous driving, especially under unfavorable ambient conditions. Various sensors have been employed to realize road area recognition under such conditions. Among these sensors, multi-channel LiDAR has recently attracted attention as it is robust against illumination changes and can obtain accurate distance information. However, this system requires a large amount of data, rendering real-time processing difficult. To overcome this limitation, feature extraction methods have been used that can extract meaningful implicit information from a large amount of data. However, the application of feature extraction to small areas of the road for real-time processing can result in reduced accuracy. To enhance the level of accuracy, an algorithm was adopted in this study that can apply feature extraction to large areas of the road. This algorithm disassembles the LiDAR data for prompt processing at a low level and then reassembles the data to manage the exceptional parts in each scan line. Subsequently, the algorithm merges the road area parts in each scan line and distinguishes the road areas, sidewalks, and obstacles in real-time. To realize these three computational processes, the algorithm is termed a disassemble-reassemble-merge (DRM) algorithm. The DRM algorithm is unique as it only uses the LiDAR sensor, which differs from existing algorithms that employ LiDAR sensors for road boundary detection. To demonstrate the effectiveness of the proposed method, we conducted an experiment using the KITTI Road Benchmark authorized dataset. Using the proposed method, accuracy was 10% higher compared to other algorithms. To validate the developed algorithm and its utility, we applied it to the Complex Yolo and obstacle-dependent Gaussian (ODG) algorithm. The ODG risk maps, which can eventually be extended to the control part of subsequent studies, were derived by applying the ODG algorithm to the outcomes and could be applied as a guideline map for actual vehicle driving. |
---|---|
ISSN: | 1598-6446 2005-4092 |
DOI: | 10.1007/s12555-022-0187-4 |