27 January 2021 in IEEE Access
The primary goal of the paper is to explore the human-vehicle-road interaction mechanism in the traffic environment and evaluate the traffic environment complexity for unmanned vehicles in urban roads. In particular, we propose the quantitative evaluation models of the traffic environment complexity for unmanned vehicles in urban roads in the paper. Specifically, the structure system of the complex traffic environment in urban roads is dissected from the aspect of human-vehicle-road, laying the basis for proposing influencing factors of traffic environment complexity. We divide the complex traffic environment into the static traffic environment and the dynamic traffic environment in light of relative static and dynamic characteristics of various environmental elements. For the complexity of the static traffic environment, the quantitative evaluation model is established by the grey relation analysis method that converts static environment complexity into the relation degree of static complexity's influencing factors. For the complexity of the dynamic traffic environment, the quantitative evaluation model is established based on the improved gravitation model that introduces the concepts of equivalent mass and the contribution degree of the unmanned vehicles' driving strategy. Besides, we evaluate the traffic environment complexity in the designed scenario by quantitative models proposed in the paper and existing evaluation models of traffic environment complexity in urban roads. The calculating process and results show that the proposed quantitative models of traffic environment complexity are more convenient and more reasonable, which provide a new idea and a method to evaluate the traffic environment complexity.
13 May 2022 in IEEE Access
It has always been an important and arduous task to detect the complex dynamic traffic environment, especially for unmanned driving. Although the existing advanced detection models have reached the speed requirements for detection, the detection accuracy needs to be further elevated to improve the unmanned driving’s safety. How to balance the accuracy and speed of detecting the complex dynamic traffic environment is still the primary problem to be solved for unmanned vehicles. Therefore, this article proposes a detection model of the complex dynamic traffic environment for unmanned vehicles by following the framework idea of YOLOv3. Firstly, we regard MobileNetv3 as the backbone and replace the traditional convolution with the depthwise separable convolution in the whole model to reduce the number of parameters and calculations. Secondly, in enhanced feature fusion layers, we perform the multi-scale fusion of four feature maps by the compress-and-expand module, the SPP module, and the cross-layer bidirectional module of feature fusion to improve the locating accuracy and reduce false detections. Thirdly, we add an IoU loss to improve the accuracy of model regression. Then, we employ the improved clustering algorithm to re-cluster anchor boxes, reducing the time overhead while improving the clustering accuracy. Finally, we compare the proposed model with other advanced detection models in the processed BDD dataset and the KITTI dataset. We verify that the mAP of the proposed model improves notably without loss of detection speed, the number of parameters and calculations decreases dramatically, and the proposed model exhibits a more superior performance.