Research prototype
Traffic Agent
A research prototype that brings transport data queries, map interaction, knowledge retrieval, and analysis reporting into one conversational workspace.
Overview
Traffic Agent is a research prototype for transport analysis tasks. It brings data queries, map interaction, knowledge retrieval, and report drafting into one conversational workspace. Users can start with a specific transport question, inspect the relevant data and spatial evidence, and retain intermediate results during a multi-turn task.
The Agent selects data, spatial-analysis, retrieval, and visualization tools according to the question. Its outputs are intended to support analysis and review rather than replace domain judgement.
Working process
Task planning
Clarify the question, the evidence needed, and the expected output.
Data queries
Select relevant datasets and inspect their definitions before analysis.
Map interaction
Run spatial checks and present results on an interactive map.
Knowledge retrieval
Retrieve supporting material when the task needs additional context.
Analysis report
Organize findings, sources, and intermediate results for review.
What it can help with
- Transport data queries
- Connect natural-language questions with relevant traffic data and query tools.
- Spatial analysis
- Inspect stations, routes, buffers, and other spatial relationships on a map.
- Knowledge and reports
- Combine retrieved context with analysis results and prepare a concise report.
Example analyses
