Abstract: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.

A Two-stage Ground Segmentation and Multi-frame Point Clouds Based Road Boundary Extraction Algorithm

Graph-based ground segmentation of 3D LIDAR in rough area

A Robust Road Segmentation Method

Road Boundary Extraction Method from Mobile Laser Scanning Point Clouds

Unstructured Road Segmentation Based on Road Boundary Enhancement Point-Cylinder Network Using LiDAR Sensor

3D Road Boundary Extraction Based on Machine Learning Strategy Using LiDAR and Image-Derived MMS Point Clouds

A Ground Segmentation Method Based on Point Cloud Map for Unstructured Roads

Real-Time Field Road Freespace Extraction for Agricultural Machinery Autonomous Driving Based on LiDAR

Automatic Road Marking Extraction and Vectorization from Vehicle-Borne Laser Scanning Data

A Staged Real-Time Ground Segmentation Algorithm of 3D LiDAR Point Cloud

Extraction of Straight Field Roads Between Farmlands Based on Agricultural Vehicle -Mounted LiDAR

Road Boundary Detection Using Multi-channel LiDAR Based on Disassemble-reassemble-merge Algorithm for Autonomous Driving

LR-Seg: A Ground Segmentation Method for Low-Resolution LiDAR Point Clouds

AUTOMATIC EXTRACTION AND RECOGNITION OF ROAD MARKINGS BASED ON VEHICLE LASER POINT CLOUD

Automatic Background Construction and Object Detection Based on Roadside LiDAR

Automatic Ground Points Identification Method for Roadside LiDAR Data

3D lidar point-cloud projection operator and transfer machine learning for effective road surface features detection and segmentation

GPS trajectory-based segmentation and multi-filter-based extraction of expressway curbs and markings from mobile laser scanning data

Road Markings Segmentation from LIDAR Point Clouds using Reflectivity Information

Automatic Extraction of Structural and Non-Structural Road Edges from Mobile Laser Scanning Data

A Fast Point Cloud Ground Segmentation Approach Based on Coarse-To-Fine Markov Random Field