A method to promote safe cycling powered by large language models and AI agents

Area of Science:

• Urban planning and transportation safety
• Artificial Intelligence and Geospatial Data Analysis
• Human-Computer Interaction for Mobility

Background:

• Urban cycling safety is a growing concern, requiring better data analysis and user-friendly tools.
• Existing methods for analyzing urban mobility and risk data are often complex and inaccessible to end-users.
• Open geospatial data offers rich information but requires sophisticated processing for practical applications.

Purpose of the Study:

• To develop a novel methodology for generating actionable information to enhance urban cycling safety.
• To integrate Large Language Models (LLMs) and AI agents with open geospatial data for improved safety insights.
• To create accessible assistive systems for cyclists and urban planners.

Main Methods:

• A multi-layer data preprocessing pipeline leveraging open geospatial data, urban risk levels, and mobility infrastructure.
• Integration of Large Language Models (LLMs) and AI-based agents for data processing and information generation.
• A defined processing pipeline encompassing Data Ingestion and Preparation, Agents Orchestration, and Decision Execution.

Main Results:

• Demonstrated a novel method combining LLMs and AI agents for processing multi-domain open geospatial data to promote cycling safety.
• Successfully integrated urban risk data and cycling infrastructure into accessible information formats (textual/audio prompts).
• Developed a proof-of-concept AI agent implementing the proposed methodology.

Conclusions:

• The proposed methodology effectively enhances cycling safety and supports urban mobility planning.
• The integration of LLMs and AI agents with open geospatial data provides a reproducible and accessible approach.
• User-friendly assistive systems can be created to provide valuable insights for safer urban cycling.