TrafficRAG for Transportation 5.0: Building “6S” Parallel Transportation Systems via Retrieval-Augmented Generation

The growing demand for personalized mobility places higher expectations on transportation systems to interact with users and accurately interpret their intentions. This necessitates the development of a digital assistant to provide seamless navigation to intelligent transportation tools and information based on the adequate comprehension of users’ needs. The emergence of foundation models makes it possible to create such a digital assistant. To this end, this article proposes a parallel transportation framework (TrafficRAG) based on foundation models and retrieval-augmented generation (RAG) to provide a unified solution to various traffic issues. In the framework, RAG is applied to overcome the limitations of foundation models, such as limited and hard-to-update knowledge as well as the “hallucinations” problem by integrating external knowledge bases. Digital humans (DHs) are built to engage naturally with biological humans and work alongside robotic humans to offer safe, reliable, and personalized services. Although the journey toward realizing TrafficRAG is still fraught with challenges, we believe that it paves a promising path to the future of Transportation 5.0.