This study is to analyze the influence of visibility in a foggy weather environment on the accuracy of machine vision obstacle detection in assisted driving. We present a foggy day imaging model and analyze the image characteristics, then we set up the faster region convolutional neural network (Faster R-CNN) as the basic network for target detection in the simulation experiment and use Karlsruhe Institute of Technology and Toyota Technological Institute (KITTI) data for network detection and classification training. PreScan software is used to build weather and traffic scenes based on a foggy imaging model, and we study object detection of machine vision in four types of weather condition—clear (no fog), light fog, medium fog, and heavy fog—by simulation experiment. The experimental results show that the detection recall is 91.55%, 85.21%, 72.54~64.79%, and 57.75% respectively in no fog, light fog, medium fog, and heavy fog environments. Then we used real scenes in medium fog and heavy fog environment to verify the simulation experiment. Through this study, we can determine the influence of bad weather on the detection results of machine vision, and hence we can improve the safety of assisted driving through further research.
17 May 2021 in IEEE Transactions on Transportation Electrification
Driving obstacle detection has great importance in automatic driving safety. However, the deterioration of image quality in the foggy environment brings many difficulties to the detection and even endangers driving safety. Therefore, in this article, a feature fusion method is designed to improve the performance of the camera sensor in driving obstacle detection under foggy weather. The difference in image features of the same traffic scene on sunny and foggy days is compared, and the reason for the decrease in detection performance of the camera sensor in foggy environments is found. The foggy image dataset is constructed, and the feature fusion of the detection object in foggy and sunny environments is realized through YOLOv3. Compared with the regular trained model, the method in this article improves 30% in mean average precision (mAP), 40% in recall, and 22% in F1-Score of the object detection performance in foggy images. Unlike the image defogging method, foggy images can be directly used for detection and have higher detection accuracy and time-efficient. This is of great significance for improving driving obstacle detection in adverse weather, which is important to ensure driving safety.
11 July 2024 in IEEE Transactions on Intelligent Vehicles
The perception system based on roadside LiDAR is an important component for autonomous driving. However, previous studies optimizing the deployment configuration of roadside LiDAR systems were constrained by LiDAR designs that failed to provide uniform coverage of vehicle point clouds within the perception area. Our objective is to enhance the perception capability of roadside sensing systems by considering both the inherent properties of LiDAR (e.g., beam distribution) and its relationship to roadside infrastructure, particularly in terms of placement height. Firstly, we proposed an analytical model to establish a quantitative relationship between LiDAR configurations and the distribution of vehicle point clouds. We conducted a field test to generate real vehicle point clouds and validate the proposed analytical model using three types of LiDARs. Secondly, we optimized the roadside LiDAR configurations to maximize the perception entropy of vehicle point clouds. We designed a particle swarm optimization algorithm using initial configurations generated through the bisection method as prior knowledge. The use of a particle swarm optimization algorithm narrowed down the search area, thereby expediting the optimization process. A set of simulation-based experiments was conducted using the NGSIM I-80 dataset. Compared to three alternative configurations, the proposed joint optimization model showed the most significant improvement for the 16-beam and 32-beam LiDAR models, with detection Recall increasing by 65% and 63%, respectively. These improvements are 4.8 times and 3 times higher than those of the base configuration. This study provides an effective method for LiDAR manufacturers and traffic professionals to optimize roadside LiDAR systems, thereby enhancing vehicle perception capabilities.
22 September 2025 in IEEE Internet of Things Journal
Developing an effective evaluation metric is crucial for accurately and swiftly measuring LiDAR perception performance. One major issue is the lack of metrics that can simultaneously generate fast and accurate evaluations based on either object detection or point cloud data. In this study, we propose a novel LiDAR perception entropy metric based on the probability of vehicle grid occupancy. This metric reflects the influence of point cloud distribution on vehicle detection performance. Based on this, we develop a LiDAR deployment optimization model, which is solved using a differential evolution-based particle swarm optimization (DE-PSO) algorithm. A comparative experiment demonstrated that the proposed PE-VGOP offers a correlation of more than 0.98 with the vehicle detection results in evaluating LiDAR perception performance. Furthermore, compared to base deployments, field experiments indicate that the proposed optimization model can significantly enhance the perception performance of various types of LiDARs, including RS-16, RS-32, and RS-80. Notably, it achieves a 25% increase in detection Recall for the RS-32 LiDAR. Additionally, sensitivity analysis under varying traffic densities further verifies the robustness of the proposed model. This study provides a practical and generalizable framework for enhancing roadside LiDAR deployment in diverse traffic environments.