đ€ AI Summary
Traditional vehicle routing approaches struggle to model the semantic, multi-step tasks and dynamic contextual constraints inherent in human drivingâsuch as urgent requests or user preferencesâwhile their multi-objective optimization relies on labor-intensive parameter tuning and lacks interpretability. This paper proposes PAVe, the first framework that tightly integrates large language model (LLM) agents with the classical multi-objective Dijkstra algorithm: the LLM handles intent parsing, hierarchical task decomposition, and dynamic constraint reasoning, while the pathfinding algorithm generates candidate routes; integration with a city-scale POI geocached knowledge base enables context-aware decision-making. PAVe thus shifts the paradigm from âshortest-pathâ to âintent-driven routing.â Evaluated on real-world urban benchmarks, it achieves an 88.3% accuracy in initial route selection, significantly enhancing both personalization and explainability.
đ Abstract
Traditional vehicle routing systems efficiently optimize singular metrics like time or distance, and when considering multiple metrics, they need more processes to optimize . However, they lack the capability to interpret and integrate the complex, semantic, and dynamic contexts of human drivers, such as multi-step tasks, situational constraints, or urgent needs. This paper introduces and evaluates PAVe (Personalized Agentic Vehicular Routing), a hybrid agentic assistant designed to augment classical pathfinding algorithms with contextual reasoning. Our approach employs a Large Language Model (LLM) agent that operates on a candidate set of routes generated by a multi-objective (time, CO2) Dijkstra algorithm. The agent evaluates these options against user-provided tasks, preferences, and avoidance rules by leveraging a pre-processed geospatial cache of urban Points of Interest (POIs). In a benchmark of realistic urban scenarios, PAVe successfully used complex user intent into appropriate route modifications, achieving over 88% accuracy in its initial route selections with a local model. We conclude that combining classical routing algorithms with an LLM-based semantic reasoning layer is a robust and effective approach for creating personalized, adaptive, and scalable solutions for urban mobility optimization.