Skip to main navigation Skip to search Skip to main content

Leveraging Large Language Models (LLMs) and generative AI in Understanding Travel Behavior Across Demographics: A Comparative Study of Different Groups Using Street View Imagery and Environment Perception

  • Chung, H. (Supervisor)
  • Richard Dunning (Co-supervisor)
  • Kon Kim (Co-supervisor)
  • Yonto, D. (Co-supervisor)

Activity: Supervision › PhD Supervision

Description

The understanding of travel behavior is of great significance, as it enables decision-makers to predict traffic flow, plan actions, and implement urban management. Large language models (LLMs) have achieved remarkable success in natural language processing. However, their applications in analyzing street view images are still relatively limited, and scenarios where generative AI are used to form new predictions and advices are even rarer.
This research will leverage the advantages of LLMs as well as generative AI to capture the relations between time periods and locations in texts, street view images, individual’s profile and travel trajectory, to better predict travel behavior, make travel plans and give advices. The research will focus on different groups of people: children, students, commuters, couriers and elderly people, which cover most of the age groups and travel modes of the entire population. This project will evaluate the performance of the model and discuss the integration of LLMs with integrate Retrieval Augmented Generation (RAG) and AI Agent and understand urban traffic patterns and personal travel behaviors.
Period15 Jan 2026 → 14 Jan 2030
ExamineeZicheng Wang
Degree of RecognitionInternational