IET Intelligent Transport Systems最新文献

Recent developments in intelligent transportation systems underscore the promise of combining deep reinforcement learning (DRL)-based traffic signal control (TSC) with automated vehicles (AVs) to improve intersection management. This study analyses how integrating DRL-based TSC systems with AVs affects traffic efficiency, safety and emissions under varying demand levels. By simulating realistic driving behaviours and using sophisticated statistical methods, the research finds that DRL-based TSC significantly outperforms traditional fixed-time and actuated systems, effectively reducing congestion, emissions and conflicts. Queue length analyses reveal that DRL-based TSC provides substantial efficiency gains, further enhanced by AVs, which reduce congestion through improved driving automation. Notably, the short-term benefits of DRL-based TSC at low AV market penetration rates resemble the long-term effects of conventional systems at high AV adoption. While fuel consumption improvements under low demand are modest compared to other adaptive systems, high-demand scenarios show significant advantages of DRL-based TSC, with AV integration further optimising flow and reducing stop-and-go patterns. Safety analysis indicates that DRL-based TSC improves intersection safety, particularly at low AV penetration, with AVs dramatically reducing conflicts. Overall, combining DRL-based TSC with AV technology holds considerable potential for advancing traffic management, safety and environmental outcomes in urban settings.

Automatic license plate recognition (ALPR) is a frequent research topic due to its wide-ranging practical applications. While recent studies use synthetic images to improve license plate recognition (LPR) results, there remain several limitations in these efforts. This work addresses these constraints by comprehensively exploring the integration of real and synthetic data to enhance LPR performance. We subject 16 optical character recognition (OCR) models to a benchmarking process involving 12 public datasets acquired from various regions. Several key findings emerge from our investigation. Primarily, the massive incorporation of synthetic data substantially boosts model performance in both intra- and cross-dataset scenarios. We examine three distinct methodologies for generating synthetic data: template-based generation, character permutation, and utilizing a generative adversarial network (GAN) model, each contributing significantly to performance enhancement. The combined use of these methodologies demonstrates a notable synergistic effect, leading to end-to-end results that surpass those reached by state-of-the-art methods and established commercial systems. Our experiments also underscore the efficacy of synthetic data in mitigating challenges posed by limited training data, enabling remarkable results to be achieved even with small fractions of the original training data. Finally, we investigate the trade-off between accuracy and speed among different models, identifying those that strike the optimal balance in each intra-dataset and cross-dataset settings.

The rapid advancement of electric vehicles (EVs) is hindered by their limited driving range. Intelligent path planning can significantly improve energy efficiency and extend driving range. This paper proposes a novel three-dimensional multi-objective path planning method considering the characteristics of the electric drive system (EDS-3DM). First, vehicle dynamics and energy consumption estimation models are developed based on the efficiency analysis of the EDS. Next, a comprehensive path evaluation model is designed using both Euclidean distance and energy consumption. B-spline curves are then applied to smooth the final paths. Experimental results on three different maps demonstrate the effectiveness of EDS-3DM, achieving an average energy consumption reduction of 12.74%.To address the path planning challenge in intelligent EVs, this paper proposes a novel three-dimensional multi-objective path planning method that considers the characteristics of the EDS-3DM.The path planning results on three maps demonstrate the effectiveness of the EDS-3DM and its ability to achieve an average energy consumption optimization of 12.74%.

This paper presents a statistical analysis of the impact of front brake lights (FBL) used in real road traffic on road safety from the perspective of the participating drivers. In contrast to traditional brake lights mounted on the rear of vehicles, the FBL provides additional information about the driver's intention to stop, especially to road traffic users looking at the front of the vehicle (e.g., when the vehicle is approaching). This innovative solution is designed to enhance road safety by offering supplementary visual cues, particularly in scenarios when it may be challenging to discern rear brake lights. In this study, 2,476 surveys were collected from drivers (both professionals and non-professionals) and analysed to determine how the presence of FBL affected their perception of road safety. The statistical investigation revealed that only 13% of participants stated that FBL had never assisted in mitigating or minimising the risk of collision. It is noteworthy that the older generation and women drivers (both professional and non-professional) evaluated FBL more positively. On the other hand, professional drivers demonstrated more scepticism and a neutral attitude towards the benefits of FBL. These findings highlight the need for targeted information campaigns.

Existing studies on trajectory optimization for cooperative automated driving systems (C-ADS) equipped vehicles at signalized intersections operate under a simplified assumption of cooperative behaviour: all vehicles accept and follow to the prescribed plans. To investigate trajectory optimization for C-ADS-equipped vehicles with different cooperation classes, a deep deterministic policy gradient (DDPG) algorithm was developed within a reinforcement learning (RL) framework, alongside baseline implementations of trajectory smoothing (TS)-based C-ADS systems and human-driven vehicle scenarios. Experimental results indicate that the proposed methodology achieves significant reductions in average travel time (53.59%) and stop times, compared to benchmark approaches. Furthermore, novel insights into the performance improvements at signalized intersections were derived from analysing different cooperation classes of C-ADS-equipped vehicles via the RL model, providing critical guidance for refining control strategies in cooperative automated driving systems. This study validates that RL models utilizing the DDPG algorithm serve as effective tools for enhancing the performance of cooperative automated driving systems.

Unmanned aerial vehicles (UAVs) are becoming integral to time-sensitive logistics and intelligent mobility systems due to their flexibility, low deployment cost, and real-time connectivity. However, their open and dynamic communication environment—typically organized as flying ad hoc networks (FANETs)—makes them highly vulnerable to a wide spectrum of cyber threats. To address this, we propose a novel multihop intrusion node detection scheme (MINDS) powered by an AI-driven ensemble learning model, X-CID, optimized for lightweight drone networks. The proposed system integrates a decentralized multi-hop architecture with intra- and inter-cluster communication validation, enabling real-time anomaly detection across the physical, communication, and architectural layers of UAV systems. To improve detection performance under resource constraints, feature selection is applied using the Pearson correlation coefficient (PCC), and model hyperparameters are fine-tuned using randomized search cross-validation. Trained and evaluated on three benchmark datasets (WSN-DS, NSL-KDD, CICIDS2017) covering 24 distinct attack types, X-CID outperforms traditional models in F1-score (up to 99.84%), accuracy (up to 99.70%), and achieves low false alarm rates with competitive latency. The proposed approach ensures robust, scalable, and energy-efficient security for autonomous drone communication, making it suitable for critical missions in logistics, disaster response, and aerial surveillance.

In recent years, multi-agent reinforcement learning (MARL) has been increasingly applied in training cooperative decision models for connected autonomous vehicles (CAVs). Despite the success they have demonstrated, they are bound to inherit issues that deep learning models suffer, such as vulnerability to adversarial attacks which is the focus of this study. Consequently, this paper aims to assess and enhance the robustness of MARL-trained cooperative policies used by CAVs, in terms of their resilience to adversarial behavior encountered during deployment. First, a specific existing cooperative policy was identified to be the victim policy, deployed in an on-ramp merging road scenario. Second, two adversarial policies, namely collision adversary ($��a�d�v_c�$) and speed adversary ($��a�d�v_s�$), were developed and trained to disrupt the performance of the victim policy. The adversarial policies significantly impacted the victim policy, increasing the collision rate to 62% and decreasing the average speed from 25 m/s to 21.73 m/s. Finally, several adversarial training approaches were developed, producing more robust cooperative policies against adversarial scenarios, by significantly bolstering road safety in adversarial conditions. The collision rate was cut by half against $��a�d�v_c�$, whereas, 0% collision scored in the face of $��a�d�v_s�$.

It is a critical problem to improve battery energy management for electric vehicle platooning systems. Moreover, different from internal combustion engine vehicles, regenerating braking is widely used to recover part of the energy in the electric vehicle when it is braking. This paper presents the optimization method of battery energy management for electric vehicle platooning with regenerating braking. By investigating the force analysis of platooning and the battery model, a new optimization strategy is presented to minimize the cost of the battery for both charging and maintaining. The cost of the battery is not only related to the state of charge (SoC) but also concerned with the state of health (SoH) due to the battery aging phenomenon. Thus, a new cost function concerned with SoC and SoH consumption is presented. Further, the optimization problem is addressed by the dynamic programming method combined with the successive convex approximation method. Finally, it is discussed how to choose the trade-off weights to adapt to different actual situations, and simulation results are provided to verify the effectiveness and advantages of the proposed methods.

Vehicular edge computing (VEC) has emerged as a promising paradigm to reduce the latency of vehicular tasks by leveraging edge computing resources. However, the high mobility of vehicles and the limited computational capacity of edge servers (ESs) present significant challenges to achieving efficient VEC. To address these challenges, this paper proposes a fine-grained computation task cooperative offloading and migration strategy. Specifically, applications are decomposed into multiple interdependent subtasks, which are collaboratively executed across multiple ESs. As vehicles move, computation tasks are dynamically migrated among ESs to ensure service continuity. The joint optimisation of task offloading and migration is formulated as a multi-stage mixed integer non-linear programming problem. To tackle this problem, we first employ Lyapunov optimisation to transform the multi-stage problem into a deterministic optimisation problem at each time slot, aiming to maximise long -term system revenue. Furthermore, considering the dynamic environment characterised by vehicle mobility, time-varying channels, subtask dependencies and inter-vehicle channel interference, we integrate a graph convolutional network (GCN) into the counterfactual multi-agent policy gradients (COMA) framework. By integrating Lyapunov optimisation with COMA-GCN, we propose Ly-COMA, a novel algorithm that effectively minimises the average task execution delay. Extensive experimental results demonstrate that the proposed algorithm outperforms existing methods in terms of average delay reduction and migration cost efficiency.

This paper presents a comprehensive, data-driven framework for personalised driving risk assessment, designed to enhance driver safety within intelligent transportation systems. By leveraging crowdsensed telemetric and road environment data, the framework captures diverse driving behaviours and contextual factors to provide real-time, individualised risk insights. The two-phase framework combines Gaussian Mixture Model (GMM) clustering, Deep Embedded Clustering (DEC), and Fully Connected Network (FCN) for accurate risk classification and prediction, while Deep Q-Learning (DQN) delivers adaptive feedback that encourages safer driving practices. Extensive evaluation shows that our approach outperforms traditional models in both accuracy and adaptability with an accuracy score of 95% and an average F1-score of 0.94, demonstrating its value in capturing complex driver behaviour patterns and contributing a scalable solution for transportation safety.