[2606.12657] TrajGenAgent: A Hierarchical LLM Agent for Human Mobility Trajectory Generation
Researchers introduced TrajGenAgent, a new framework for generating realistic human mobility trajectories without fine-tuning models. This system uses a hierarchical approach to synthesize individual activity chains and ground them into realistic visit patterns. Experiments show TrajGenAgent outperforms existing models in spatiotemporal fidelity and behavioral realism while maintaining computational efficiency.
This is worth holding only if the practical relevance is clear from the source.
This record is extracted from a published AI Today issue and tied to the original source URL. Treat the source as the record of evidence for the summary.