TrajGenAgent uses hierarchical LLM planning for synthetic mobility data

TrajGenAgent uses a hierarchical LLM-agent design to generate human mobility trajectories.

The system separates high-level planning from more detailed trajectory construction, aiming to preserve reasoning while adding spatiotemporal structure. That matters because mobility data is valuable but often costly and privacy-sensitive to collect.

Synthetic trajectory generation can support transportation planning, urban modeling, and epidemic analysis if the generated data is realistic enough to be useful.