Real-time traffic signal control represents a key challenge in modern intelligent transportation systems, particularly under highly variable traffic flows and the presence of priority vehicles. This study proposes a hybrid framework for adaptive signal plan control at a signalized intersection. The framework integrates deep learning-based traffic prediction, surrogate-based performance evaluation, and reinforcement learning-based adaptive control. Short-term traffic flow is predicted using recurrent neural networks, providing anticipatory information for traffic control decisions. Based on predicted flows and generated candidate signal plans, a machine learning surrogate model enables fast estimation of key performance indicators, including average vehicle delay and queue length. Adaptive control is implemented using the Proximal Policy Optimization algorithm within the SUMO environment via TraCI, which enables real-time fine-tuning of signal phases. A dedicated priority and stability module ensures effective emergency vehicle preemption and adaptive public transport priority while preserving intersection stability. Simulation results show that the proposed framework reduces average vehicle delay by up to 35% compared with FT and by up to 15% compared with standalone RL, while also improving traffic flow efficiency and priority vehicle performance.

Traffic signal control at urban intersections is one of the key determinants of the overall efficiency of the transportation system, given its direct impact on travel time, congestion levels, and emissions of exhaust fumes. This study proposes an integrated hybrid model that combines a metaheuristic Genetic Algorithm for generating potential signal timing plans with fuzzy multi-criteria decision-making (MCDM) for their evaluation and selection of the optimal solution. In order to determine the relative importance of criteria, the fuzzy methods F-AHP, F-FUCOM, and F-PIPRECIA were employed, thus providing stable assessments of criteria importance under conditions of uncertainty and expert subjectivity. The ranking of generated alternatives was performed by employing the F-TOPSIS, F-WASPAS, and F-ARAS methods, while the robust decision-making rule approach was employed to develop a robust decision-making rule by integrating multiple MCDM methods. The proposed model was tested using data collected from a real urban intersection. The results show that the integrated hybrid approach enables a significantly more reliable selection of the optimal signal timing plan and achieves higher traffic management efficiency compared to traditional methods. The proposed model provides a flexible and scalable framework that can be adapted to different types of intersections and traffic demand conditions, thereby significantly contributing to the development of modern intelligent traffic management systems.

Artificial Intelligence (AI) is increasingly applied to road intersection management to optimize traffic flow, reduce congestion, and improve safety. However, the opaque nature of AI-based decisions can undermine trust and limit adoption in intelligent transportation systems. This paper extends research on AI-driven intersection management by integrating Explainable Artificial Intelligence (XAI) methods ensuring the decisions of AI are interpretable and trustworthy for real-world deployment by traffic management authorities. A YOLO-based model is used for vehicle detection and traffic state recognition, while XAI techniques are applied to provide interpretable insights into system decisions. Implementation on the NVIDIA Edge AI device is considered and tested on a large dataset of real-world road scenes with multiple traffic objects, including various vehicle types and pedestrians. Results from experiments show that XAI integration enhances system comprehensibility without significantly reducing performance. These findings underline the importance of explainability for deploying trustworthy, accountable, and safe AI solutions in road intersection management.