Controllable Trajectory Generation

I explore generative models for creating realistic and controllable trajectories with various semantic constraints.

Key Works

• ControlTraj (KDD 2024): Controllable trajectory generation with topology-constrained diffusion model
• DiffTraj (NeurIPS 2023): Generating GPS trajectories with diffusion probabilistic model
• SynMob (NeurIPS 2023): Creating high-fidelity synthetic GPS trajectory datasets

Research Focus

• Develop topology-constrained diffusion models for realistic trajectory generation
• Enable conditional generation with semantic constraints
• Create synthetic datasets for data augmentation and privacy protection