Traffic Congestion Control with Emergency Awareness and Optimized Communication Infrastructure using Reinforcement Learning and Non-Dominated Sorting Genetic Algorithm

Urban centers are grappling with increasing traffic congestion, which hampers mobility and reduces the effectiveness of emergency responses. Current traffic management systems struggle to dynamically prioritize emergency vehicles during peak congestion, highlighting the need for an advanced solution that improves both everyday traffic flow and emergency response efficiency. Traditional traffic light systems typically follow either fixed schedules (“Fixed” approach) or make real-time adjustments without achieving long-term efficiency (“Greedy” approach), often failing to effectively balance regular traffic dynamics with emergency needs. We propose the “Avg Log“ method, an adaptive Q-learning strategy that enhances conventional traffic management systems by incorporating an emergency-aware module. This module uses real-time data to prioritize emergency vehicles and adjust traffic signals accordingly. Additionally, our approach employs a multi-objective optimization algorithm to strengthen communication infrastructure, optimizing latency, throughput, and reliability. Through simulations in a multi-intersection urban setting, we compared the performance of our method against the standard “Greedy” and “Fixed” approaches. Key metrics assessed included average waiting times, system throughput, and the efficiency of emergency vehicle passage. The “Avg Log” method outperformed traditional methods, showing superior long-term adaptability and learning. It demonstrated lower average waiting times and effectively managed emergency situations, significantly reducing delays for emergency vehicles compared to both control methods. This method provides an efficient balance between overall traffic flow and emergency prioritization, suggesting a scalable solution for complex urban environments.