Driver’s attention area recognition (DAAR) is crucial for autonomous vehicle takeover and detecting distracted driving. Currently, deep learning-based models for DAAR have achieved promising results. However, due to substantial variations in driving environments and individual driver characteristics, end-to-end deep learning models struggle to maintain stable generalization performance across different driving scenarios. To address these challenges, this paper proposes a driver attention area recognition framework based on improved Transformer and Kalman Filtering (DAARF-TK). DAARF-TK comprises two key components: a facial landmarks extraction module and a Transformer modeling module. The face landmarks extraction module includes an Adaptive Unscented Kalman Face Landmark Filtering (AUKFLF) algorithm, calibrating face landmarks in scenarios with large face occlusions, significant head rotations, and blurriness. An efficient attention mechanism and a Tokens Dimension Reduction Module are embedded into the Transformer to enhance the adaptability of the Transformer architecture to the real-time requirements of the DAAR task. Experimental results reveal that when trained on the Recorded dataset, DAARF-TK attains mAcc values of 94.13%, 91.67%, and 75.38% on the Recorded, 100-Driver, and DGW test sets, respectively. In addition, a real-time validation conducted on a driving simulator indicates that DAARF-TK operates at 37.68 frames per second (FPS), confirming its real-time capability in DAAR. These findings underscore the superiority of DAARF-TK and its potential to offer technical support for autonomous vehicle takeover.

Connected and autonomous vehicles (CAVs) are emerging as a potential solution to merging safety problems. However, in mixed traffic scenarios where CAVs coexist with human-driven vehicles (HVs), challenges arise due to the lack of proactive cooperation and the limited length of the acceleration lane, complicating the merging processes of CAVs. In these cases, CAVs should actively seize the transit opportunity and perform safe and efficient merging. Failure to do so can lead to decreased traffic efficiency, increased fuel consumption and emissions, compromised self-merging capacities, and heightened crash risk. Therefore, this article employs the roadside unit and proposes a two-level hierarchical self-merging strategy for CAVs to increase the merging efficiency while ensuring high safety. Since the surrogate safety measures (SSMs) can formulate reliable safety assessment and identify the merging conflict risk by setting appropriate threshold, the upper level uses a novel SSM-based method (i.e., the minimum acceleration rate, MIAR) to determine the merging sequence (MS). A theoretical model for merging safety (TMMS) is developed to estimate the MIAR value, and the MIAR threshold is determined using signal detection theory (SDT). At the lower level, the strategy recommends the optimized merging maneuvers, pregenerated using sequential quadratic programming-model predictive control (SQP-MPC), based on the determined MS and the CAV’s velocity. A case study at a real-world freeway merging area demonstrates the effectiveness of the MIAR in measuring merging conflict risk. Besides, numerous simulations are conducted, and the results demonstrate that the proposed strategy significantly improves merging success rate and overall traffic efficiency.

The reflections of multi-colored lights involved on rainy nights present strong uncertainties and abruptness, resulting in a high rate of false and missed detections in existing methods. To solve this issue, this paper proposes a lane detection method based on memory and discretization mechanisms. Firstly, a Memory Fruit-fly-optimizer with Individual Differences (MFID) is innovatively proposed to drive Multi-threshold Otsu (MOtsu)-based multi-class segmentation of lanes, which is a high-dimensional optimization task with real-time and local optimal challenges, for capturing lane clues obscured in multi-intensity reflections, consequently reducing missed detections. Specifically, to solve the challenges inherent in the task, the MFID incorporates a novel memory mechanism to establish fast-converging initial conditions for real-time detection, while creatively considering individual differences to motivate multi-swarm optimization that mitigates local optima risks. After integration, the MFID-MOtsu is constructed for lane segmentation. Subsequently, a dynamic discretization mechanism is proposed to efficiently separate lane edges from interference edges, mitigating accuracy degradation caused by their entanglement. Finally, the false detection issue is greatly reduced through the implementation of adaptive geometric filters. The experimental results demonstrate that the proposed method achieves an average accuracy of 93.21% on rainy nights, indicating an average improvement of 12.7% over state-of-the-art methods. Additionally, without any parameter modifications, the proposed method is applicable to both normal and classic challenging scenes, such as nights, tunnels, rainy days, and shadows. The algorithm achieves an average accuracy of 96.2% and an average detection speed of 46 frames per second.

Driver distraction-level recognition while performing secondary tasks in full-touch in-vehicle information systems (FTIVISs) is essential for the harmonious co-driving of human and intelligent vehicle systems. However, there has been little research on this topic. To respond to this issue, this paper proposes a distraction-level recognition framework with a combination of semi-supervised learning, unsupervised learning, and supervised learning. First, unsupervised learning and semi-supervised is used to divide the collected unlabeled samples of driving distraction behavior into three categories of distraction levels: high, medium, and low. Second, the factors influencing the distraction level are explored through a mixed model analysis. Finally, a stacking-based ensemble learning model is proposed to recognize the driver distraction level by supervised learning, with the influencing factors of the distraction level used as model input parameters. To improve the recognition performance of the stacking model, four heterogeneous classifiers selected as the base classifiers, and principal component analysis (PCA) is introduced in the base classifier layer, which improves the ability of the meta -classifier to process raw features based on overfitting resistance. We conducted a real road experiment under different road and FTIVIS task conditions, and the proposed model performed better than traditional machine learning models. In addition, the model exhibited the greatest advantage when the deep neural network (DNN) algorithm was used as the meta -classifier of the model, with a recognition accuracy of 92.5%. The study findings are significant for developing a human–machine co-driving control strategy and improving vehicle driving safety.

In recent years, the full-touch human–machine interface (HMI) mode has been widely used in vehicles built by Tesla. This interaction mode replaces the conventional physical interaction modality with a screen, and it has a good sense of technological experience. However, it is unclear whether this mode will make the driver's lateral control more challenging than the conventional mode (CM). To investigate this issue, two most common secondary tasks were designed: dialing and navigation entry tasks and real-world road experiments were conducted using two instrumented vehicles. The vehicle operating parameters and the driver manual data were collected in different modes, respectively. Interestingly, the opposite results were found regarding the effect of the full-touch mode (FTM) on the driver's lateral control ability in different secondary tasks. Compared with the CM, the lateral control ability decreased less during the dialing task relative to the baseline driving in the FTM, while the lateral control ability decreased more in the FTM during the navigation entry task. In addition, drivers’ lateral control decreased further as task difficulty and driving speed increased regardless of mode. This study provides a theoretical basis for the development of laws and regulations regarding full-touch HMI mode.

Identifying distracted drivers is crucial for enhancing driving safety and advancing intelligent driver assistance systems. Recently, researchers have applied Convolutional Neural Network (CNN) and Vision Transformer (ViT) models for driver state decision. However, both models often suffer from several issues such as numerous parameters and low detection efficiency. To address these challenges, this study proposes the Convolution Vision Transformer (CoViT) model for distracted driver identification, leveraging techniques such as Low Complexity Attention Mechanism (LCAM), Multi-scale Dilation Convolution (MSDC), and Depth Separable Convolution (DSC). Moreover, the CoViT model features a typical “pyramid” structure, enabling effective feature extraction across different scales. Subsequently, the proposed system is trained and evaluated using the publicly available driving behavior datasets SFD2 and 100-Driver, as well as real-world road experiments. Experimental results show that the CoViT model yields high recognition performance, with mean Accuracy (mAcc) scores of 95.17%, 97.89%, and 93.54% on the recorded dataset, SFD2 dataset, and 100-Driver dataset, respectively. These scores surpass those obtained by similar lightweight models. Furthermore, ablation experiments reveal that deep and dilated convolution significantly enhance model performance. In addition, the CoViT model demonstrates its applicability to real-time driving behavior detection tasks, with a parametric count of just 1.24M — a reduction of 2.67M compared to MobileNetV3 — and an online inference Frames Per Second (FPS) of 159.13.

Real-time detection of oncoming vehicle maneuvers at intersections is essential for connected autonomous vehicles (CAVs) to plan safe paths and driving strategies. Most existing methods use supervised learning methods to construct behavior detection models and assume that most data have labels. Real data collected by the onboard sensor as a data stream is unstable, and there are outliers, concept drift, and evolution problems, potentially decreasing the detection accuracy. To this end, we propose a micro-cluster-based online adaptable (MCOA) approach. The framework consists of four parts: initial model construction, new class detection, classification using k-nearest neighbor (k-NN), and online update. First, k-means clustering is performed on the maneuvering behavior data, and cluster features are derived to obtain a set of micro-clusters (MCs) to establish the initial model. Second, we analyze the instances stored in the data block to detect new classes and use the k-NN to classify the incoming instances. Finally, the model is updated online using an update strategy based on error-driven representative learning, a time-effect function, and a local decision boundary. A driving simulator is used to collect experimental data consisting of left turns (LT), right turns (RT), and going straight (GS) to establish and evaluate the model. The results show that the proposed model achieves higher detection accuracy for early-stage intersection maneuvers and has stronger adaptability to new classes than benchmark algorithms.

Driving simulators have been widely used in driving behavior analysis and intelligent driving algorithm development. However, the validity of driving behavior data derived from driving simulators remains unclear. In this study, 30 Chinese drivers were recruited to participate in two experiments: on-road and simulator experiments. An instrumented vehicle and a high-fidelity simulator were used in the on-road and simulator experiments, respectively, to investigate the effects of high speed (60, 80, and 100 km/h) and a visual distraction task on the lateral driving performance, including lane deviation (LD), standard deviation of the lane position (SDLP) rate, standard deviation of the steering wheel angle (SDSWA) rate, and steering wheel reversal rates (SRRs) (at levels of 1.3° and 2.5°). It was found that the visual distraction task impaired the drivers’ lane-keeping ability. Furthermore, the driving task had similar effects on the LD, SDLP rate, and SRRs (2.5°) in the on-road and simulator experiments. The effects of the driving speed on the LD, SDLP rate, and SDSWA rate were comparable in both driving environments. However, the results confirmed that even a high-fidelity driving simulator could not achieve perfect absolute validity. The results provided preliminary evidence that the high-fidelity driving simulator used in this study might be an effective tool for investigating the effects of visual distractions task on lateral driving behavior.

Vehicle-to-Infrastructure (V2I) communication has provided a solution for the improvement of the traffic efficiency of smart city intersections. For example, turning maneuvers prediction at signalized intersections in a connected environment helps traffic command centers time traffic lights and dynamically predict traffic flow. However, the modeling methods used in existing research on this topic have some limitations, such as poor scalability and interpretability of machine learning. Thus, this study proposes a dictionary learning-based approach to predict turning maneuvers before the intersection. The proposed dictionary model estimates the LogDet divergence-based sparse inverse covariance matrix (LDbSICM) of driving behavior samples. The graphical lasso method is used to estimate the sparse inverse covariance matrix of the driving samples to construct a dictionary library of the maneuver behavior. The LogDet divergence is used to calculate the difference between each inverse covariance matrix. A driving simulator is utilized to collect experimental data consisting of turning left (TL), turning right (TR), and going straight (GS) behaviors to establish and evaluate the proposed model. The experimental results demonstrate that the proposed dictionary learning-based turning maneuver prediction model achieves 100% prediction accuracy for TL and GS and 97.2% for TR. The proposed model has substantial advantages over existing methods. The model can predict TL, TR, and GS in a connected environment 270, 280, and 290 m, respectively, before the intersection.

The low user acceptability of advanced driver assistance systems (ADASs) is one of the fundamental problems limiting their wider adoption. One of the key factors affecting the frequency of warning signals is the standard warning thresholds set in ADASs, which may not be compatible with all drivers owing to various driver characteristics affecting their cognition of risk, such as comfort, skill, and experience. The present study focuses on the lane change warning (LCW) system to evaluate driving styles according to the perception of lane change risk, and determine appropriate warning thresholds corresponding to the different driving styles. A theoretical lane change warning model is established to calculate the deceleration required for the rear approaching vehicle in the target lane. The results of risk assessment experiments conducted on an actual expressway using 15 proficient drivers are employed to evaluate the risk cognitions of the participants, where the participants are divided into four different driving styles according to their adaptive warning thresholds, which are denoted as very low threshold, low threshold, medium threshold, and high threshold driving styles. Signal detection theory is then employed to determine lane change warning thresholds appropriate to the different driving styles. Our results clearly demonstrate the disparate perceptions of lane change risk for the different drivers. Therefore, the LCW system should adopt different warning thresholds for different drivers according to their driving style. The findings provide evidence supporting the exploitation of driving styles for adopting different warning thresholds in the LCW system to enhance its user acceptability.