都市道路網における局所交通負荷に基づく効率的な迂回認識型動的ルーティング

We investigate a decentralized traffic routing strategy that leverages local traffic load (LTL) information to improve the overall performance of urban transportation networks. Unlike conventional approaches based on static shortest-distance paths or globally optimized travel times, the proposed strategy dynamically balances travel distance with real-time local congestion levels, enabling efficient detour decisions with minimal computational overhead. Using a two-dimensional cellular automaton model, we simulate traffic dynamics under different routing schemes and evaluate key performance indicators, including average velocity, flow, and arrival rate. The results show that the LTL-based strategy significantly enhances system throughput and stability across a broad range of traffic densities. We attribute this improvement to its ability to induce timely, locally informed detours that suppress congestion buildup and spatial heterogeneity. Further analyses of detour statistics, travel distances, and spatiotemporal vehicle density fluctuations elucidate the underlying mechanisms. The robustness of the strategy is also confirmed on real-world urban network topologies. These findings suggest that routing strategies based on localized feedback can provide scalable and adaptive solutions for mitigating congestion in complex traffic systems.