Aplicação de Modelos de Linguagem de Grande Escala e Agentes Inteligentes com SUMO para Automatização de Simulações de Tráfego Urbano

Abstract

Simulations are important for addressing urban mobility challenges in cities and the automotive industry, allowing for an understanding of transportation systems. However, the complexity of modeling urban scenarios and the volume of data involved pose some challenges, highlighting the need for automated methods. Thus, the aim of this work is to explore the capability of Large Language Models (LLMs) in automating the urban simulation process, providing an alternative to traditional methods. To achieve this goal, the integration of LLMs with the Simulation of Urban MObility (SUMO) is proposed, using the LangChain tool. This integration allows for the construction of agents capable of configuring simulations, analyzing results, and proposing adaptive solutions. A case study is conducted to validate the proposed approach, using real data from the urban area of Natal-RN city. The results highlight the importance and relevance of applying language models in conducting simulations and interpreting generated data. These findings reinforce the significance of LLMs in automating and optimizing processes, contributing to the development of more effective and sustainable solutions for smart cities.