Summary
Many tools that predict how people move in cities need detailed travel data from those cities, which is often unavailable. The authors designed a new approach, Nomad, that learns movement patterns from cities where data exists and then applies these patterns to new cities using only location types and coordinates of places. Instead of predicting exact destinations, Nomad predicts relative moves from one place to another, making it easier to transfer knowledge between different cities. Tests show this method creates more realistic travel patterns than existing techniques even without direct data from the new city.
What this means in practice
- •For urban planners: Generate realistic human movement patterns in cities with no available mobility data by using only maps of locations and categories.
- •For location-based service developers: Improve location prediction and recommendation systems in new cities without needing prior movement data, using learned travel behaviors from other cities.
Authors
Yidi Wang, Yunhe Zhang, Bangchao Deng, Dingqi Yang, Pengyang Wang
Abstract
Individual mobility trajectories support urban analysis and location-based services, yet most trajectory generators require observations from their deployment city. This assumption excludes precisely the cities where trajectories are unavailable even though points of interest (POIs) and their attributes can be obtained from public maps. We study target-trajectory-free generation: learning from POIs and trajectories in source cities while utilizing only POI coordinates and categories in a target city, with no target trajectory or trajectory-derived statistic available for training, model selection, or generation. Existing trajectory generators typically predict absolute destinations, entangling reusable movement behavior with city-specific POI identities and spatial layouts. Our core insight is to replace this city-bound output with context-conditioned relative transitions. We propose Nomad, a transfer-and-ground framework that separates learning how people move from determining where those movements are realized. Specifically, a history-conditioned flow-matching model learns from source trajectories a transition prior over semantic displacement between POI contexts, geographic displacement, and elapsed time; at inference, a behavior graph and an exploration--return walk ground sampled transitions onto the target POI map. This factorization enables a direct test of representation level transferability without assuming invariance of the full mobility distribution. Extensive experiments across ten cities and 14 transfers show that Nomad outperforms adaptation baselines in trajectory fidelity and downstream utility, lowering the average error over the best baseline of each metric by about 15% in distributional fidelity and about 3% in downstream utility.