Abstract:The stable detection and tracking of high-speed vehicles on the road by using LiDAR can input accurate information for the decision-making module and improve the driving safety of smart cars. This paper proposed a novel LiDAR-based robust vehicle detection method including three parts: point cloud clustering, bounding box fitting and point cloud recognition. Firstly, aiming at the problem of clustering quality degradation caused by the uneven distribution of LiDAR point clouds and the difference in clustering radius between point cloud clusters in traditional DBSCAN (TDBSCAN) obstacle clustering algorithms, an improved DBSCAN algorithm based on distance-adaptive clustering radius (ADBSCAN) is designed, and a point cloud KD-Tree data structure is constructed to speed up the traversal of the algorithm; meanwhile, the OPTICS algorithm is introduced to enhance the performance of the proposed algorithm. Then, by adopting different fitting strategies for vehicle contour points in various states, the adaptability of the bounding box fitting algorithm is improved; Moreover, in view of the shortcomings of the poor robustness of the L-shape algorithm, the principal component analysis method (PCA) is introduced to obtain stable bounding box fitting results. Finally, considering the time-consuming and low-accuracy training of traditional machine learning algorithms, advanced PointNet in deep learning technique is built to send the point cloud within the bounding box of a high-confidence vehicle into PointNet to complete vehicle recognition. Experiments based on our autonomous driving perception platform and the KITTI dataset prove that the proposed method can stably complete vehicle target recognition and achieve a good balance between time-consuming and accuracy.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to use LiDAR (Light Detection and Ranging) to achieve stable detection of high - speed vehicles on the road in the autonomous driving scenario. Specifically, the author aims to overcome the problems existing in point cloud clustering, bounding box fitting and point cloud recognition in traditional methods, so as to improve the accuracy and robustness of detection, thereby providing accurate information for the decision - making module and enhancing the driving safety of intelligent vehicles. ### Detailed problem description: 1. **Point cloud clustering problem**: - When the traditional DBSCAN algorithm processes LiDAR point clouds, due to the uneven distribution of point clouds and the difference in clustering radii between different point cloud clusters, the clustering quality declines. - Solution: An improved DBSCAN algorithm (ADBSCAN) based on distance - adaptive clustering radius is proposed, and the KD - Tree data structure is used to accelerate the search during the algorithm traversal stage. 2. **Bounding box fitting problem**: - The traditional L - shape algorithm performs poorly when dealing with specific situations (such as only detecting the front or rear of a vehicle), and due to the randomness of the RANSAC algorithm itself, the fitting results are unstable. - Solution: The principal component analysis (PCA) method is introduced to obtain stable bounding box fitting results, and different fitting strategies are adopted for vehicle contour points in different states. 3. **Point cloud recognition problem**: - Traditional machine learning algorithms are time - consuming in training and have low accuracy. - Solution: The PointNet model in deep learning is used to classify and recognize the point clouds of high - confidence vehicles, so as to improve the accuracy and efficiency of recognition. ### Main contributions of the paper: - A robust vehicle detection method based on LiDAR is proposed, which includes three parts: point cloud clustering, bounding box fitting and point cloud recognition. - An improved DBSCAN algorithm (ADBSCAN) is designed. By adaptively adjusting the clustering radius and introducing KD - Tree to accelerate the search, the clustering effect is improved. - The OPTICS algorithm is introduced to enhance the clustering performance and solve the clustering problem under sparse point clouds. - The PCA method and different fitting strategies are used to improve the adaptability and stability of bounding box fitting. - The PointNet model is constructed for point cloud classification, achieving efficient and accurate vehicle recognition. Through these improvements, this method can stably complete vehicle target recognition in complex environments and achieve a good balance between time and accuracy.

Robust LiDAR-Based Vehicle Detection for On-Road Autonomous Driving

Robust vehicle detection using 3D Lidar under complex urban environment

Adversarial Robustness Analysis of LiDAR-included Models in Autonomous Driving

3D Vehicle Detection Using Cheap LiDAR and Camera Sensors.

On-road multiple obstacles detection using color images and LiDAR point clouds

Real-Time And Robust 3D Object Detection with Roadside LiDARs

Automatic Background Construction and Object Detection Based on Roadside LiDAR

Embedded and Real-Time Vehicle Detection System for Challenging On-Road Scenes

Vehicle Detection and Tracking with Roadside LiDAR Using Improved ResNet18 and the Hungarian Algorithm

3D Fast Object Detection Based on Discriminant Images and Dynamic Distance Threshold Clustering

Roadside LiDAR Vehicle Detection and Tracking Using Range and Intensity Background Subtraction

Real-Time Vehicle Detection Framework Based on the Fusion of LiDAR and Camera

A Small-Object-Detection Algorithm Based on LiDAR Point-Cloud Clustering for Autonomous Vehicles

Leveraging Deep Convolutional Neural Networks Pre-Trained on Autonomous Driving Data for Vehicle Detection From Roadside LiDAR Data

RT3D: Real-Time 3-D Vehicle Detection in LiDAR Point Cloud for Autonomous Driving

Obstacle Detection for Autonomous Driving Vehicles With Multi-LiDAR Sensor Fusion

A LiDAR Multi-Object Detection Algorithm for Autonomous Driving

Robustness-Aware 3D Object Detection in Autonomous Driving: A Review and Outlook

Obstacle Detection by Autonomous Vehicles: An Adaptive Neighborhood Search Radius Clustering Approach

Robust Lane-Detection Method for Low-Speed Environments

LiDAR-based Object Detection Failure Tolerated Autonomous Driving Planning System.