Crash risk identification and prediction are expected to play an important role in traffic accident prevention. However, most of the existing studies focus only on highways, not on multi-lane weaving areas. In this paper, a potential collision risk identification and conflict prediction model based on extending Time-to-Collision-Machine Learning (TTC-ML) for multi-lane weaving zone was proposed. The model can accurately learn various features, such as vehicle operation characteristics, risk and conflict distributions, and physical zoning characteristics in the weaving area. Specifically, TTC was used to capture the collision risk severity, and ML extracted vehicle trajectory features. After normalizing and dimensionality reduction of the vehicle trajectory dataset, Naive Bayes, Logistic Regression, and Gradient Boosting Decision Tree (GBDT) models were selected for traffic conflict prediction, and the experiments showed that the GBDT model outperforms two remaining models in terms of prediction accuracy, precision, false-positive rate (FPR) and Area Under Curve (AUC). The research findings of this paper help traffic management departments develop and optimize traffic control schemes, which can be applied to Intelligent Vehicle Infrastructure Cooperative Systems (IVICS) dynamic warning.

With their advantages of high experimental safety, convenient setting of scenes, and easy extraction of control parameters, driving simulators play an increasingly important role in scientific research, such as in road traffic environment safety evaluation and driving behavior characteristics research. Meanwhile, the demand for the validation of driving simulators is increasing as its applications are promoted. In order to validate a driving simulator in a complex environment, curve road conditions with different radii are considered as experimental evaluation scenarios. To attain this, this paper analyzes the reliability and accuracy of the experimental vehicle speed of a driving simulator. Then, qualitative and quantitative analysis of the lateral deviation of the vehicle trajectory is carried out, applying the cosine similarity method. Furthermore, a data-driven method was adopted which takes the longitudinal displacement, lateral displacement, vehicle speed and steering wheel angle of the vehicle as inputs and the lateral offset as the output. Thus, a curve trajectory planning model, a more comprehensive and human-like operation, is established. Based on directional long short-term memory (Bi–LSTM) and a recurrent neural network (RNN), a multiple Bi–LSTM (Mul–Bi–LSTM) is proposed. The prediction performance of LSTM, MLP model and Mul–Bi–LSTM are compared in detail on the validation set and testing set. The results show that the Mul–Bi–LSTM model can generate a trajectory which is very similar to the driver’s curve driving and have a preferable generalization performance. Therefore, this method can solve problems which cannot be realized in real complex scenes in the simulator validation. Selecting the trajectory as the validation parameter can more comprehensively and intuitively reflect the simulator’s curve driving state. Using a speed model and trajectory model instead of a real car experiment can improve the efficiency of simulator validation and lay a foundation for the standardization of simulator validation.

1. Driving intention, characterized by ambiguity, dynamism, and unpredictability, necessitates implementing human-like thinking and effective driving intention prediction as pivotal steps toward fu...

For accurate and effective automatic vehicle identification, morphological detection and deep convolutional networks were combined to propose a method for locating and identifying vehicle models from unmanned aerial vehicle (UAV) videos. First, the region of interest of the video frame image was sketched and grey-scale processing was performed; sub-pixel-level skeleton images were generated based on the Canny edge detection results of the region of interest; then, the image skeletons were decomposed and reconstructed. Second, a combination of morphological operations and connected domain morphological features were applied for vehicle target recognition, and a deep learning image benchmark library containing 244,520 UAV video vehicle samples was constructed. Third, we improved the AlexNet model by adding convolutional layers, pooling layers, and adjusting network parameters, which we named AlexNet*. Finally, a vehicle recognition method was established based on a candidate target extraction algorithm with AlexNet*. The validation analysis revealed that AlexNet* achieved a mean $F_1$ of 85.51% for image classification, outperforming AlexNet (82.54%), LeNet (63.88%), CaffeNet (46.64%), VGG16 (16.67%), and GoogLeNet (14.38%). The mean values of $P_{cor}$, $P_{re}$, and $P_{miss}$ for cars and buses reached 94.63%, 6.87%, and 4.40%, respectively, proving that this method can effectively identify UAV video targets.