EvoRoute: An Evolutionary Multi-Agent Framework for Sustainable and Congestion-Aware Routing in Intelligent Transportation Systems

Urban transportation systems are under increasing strain from congestion, fuel inefficiency, and emissions, which not only impact commuter mobility but disrupt supply chains (SC), leading to delays and food waste in the distribution of perishable goods. These challenges highlight the need for adaptive, sustainable, and congestion-aware routing solutions in intelligent transportation systems (ITS). This paper introduces EvoRoute, an evolutionary multi-agent framework that integrates neuroevolutionary reinforcement learning, genetic policy mutation, and multi-objective evolutionary algorithms, such as NSGA-II and MOEA/D, to jointly optimize travel time, fuel consumption, and emissions. A hierarchical policy controller balances macro level city-wide routing with micro-level intersection decisions, while eco-routing strategies minimize idle emissions and fuel wastage. EvoRoute is validated through a SUMO–OpenAI Gym co-simulation using the Bengaluru traffic dataset, with Bengaluru serving as a challenging benchmark due to its extreme congestion levels. Results show over 20% improvement in throughput and 30% reduction in carbon footprint compared to DQN, GA-only, and heuristic baselines. Beyond traffic efficiency, EvoRoute demonstrates direct benefits for sustainable food logistics, reducing delays in perishable goods delivery and contributing to food waste minimization. Overall, EvoRoute provides a scalable ITS solution aligned with SDG 11, SDG 12, and SDG 13.