CityTrajBench: A Unified Benchmark for City-Scale Vehicle Trajectory Generation
Researchers have introduced CityTrajBench, a new benchmark framework designed to standardize the evaluation of urban vehicle trajectory generation methods. This framework addresses the fragmentation in existing studies by unifying data processing, feature construction, and evaluation metrics. CityTrajBench supports various generation models, including VAEs, GANs, diffusion models, and flow-matching models, and assesses them across multiple criteria such as spatial realism, trip distribution fidelity, and geometric similarity. AI
IMPACT Standardizes evaluation for urban mobility AI, enabling clearer comparisons of future trajectory generation models.