TErouting: Traffic-aware energy routing for electric vehicle logistics fleets

The evolution of dynamic traffic environments has significantly increased the complexity of route planning for electric logistics vehicle fleets. In fleet energy management optimization, deeply integrating static route planning with dynamic traffic information and real-time vehicle states has become key to overcoming existing energy efficiency bottlenecks. This paper proposes a Traffic-Evolution-Oriented Energy Routing method for electric logistics fleets (TErouting). The method is inspired by the reasoning architecture of large language models (LLM) and manually encodes similar multi-step reasoning and temporal integration behaviors into a heuristic search framework, rather than directly calling any LLM or using conversational interfaces. Within this approach, a closed-loop optimization framework—comprising dynamic traffic perception, fleet energy modeling, and cooperative path decision-making—is constructed. The framework dynamically analyzes real-time data such as road congestion indices and charging-station queuing conditions. By integrating these with each vehicle's current state of charge (SOC) and delivery time-window constraints, the system generates an initial globally energy-optimal routing plan. Furthermore, advanced mechanisms such as self-consistency reasoning and tree-of-thought strategies are introduced to enable the model to proactively anticipate the evolution of traffic congestion. Through multi-agent Coordination, it dynamically optimizes both charging schedules and route coordination, achieving intelligent resource allocation across the fleet. The proposed method effectively addresses three major limitations of conventional models—namely, insufficient fleet-level coordination, limited real-time adaptability, and weak coupling with vehicle energy characteristics. Experimental results validate the effectiveness of the proposed optimization framework, demonstrating that the TErouting strategy significantly reduces total fleet energy consumption under dynamic traffic conditions while simultaneously enhancing overall delivery efficiency.