How Do We Move: Modeling Human Movement with System Dynamics

Modeling how human moves in the space is useful for policy-making transportation, public safety, and health. The movements can be viewed as a dynamic process that transits between states (e.g., locations) over time. In world where intelligent agents like humans or vehicles with drivers play an important role, of mostly describe activities, state transition influenced by both decisions physical constraints from real-world system need to spend time move certain distance). Therefore, modeling should include agent's decision dynamics. this paper, we propose MoveSD model movement novel perspective, learning integrating learns Generative Adversarial Imitation Learning integrates stochastic dynamics process. To best our knowledge, are first learn moving extensive experiments on datasets, demonstrate proposed method generate trajectories similar ones, outperform state-of-the-art methods predicting next location generating long-term future trajectories.