(This paper has been published in the 2019 International Conference on Computer Animation and Social Agents.)
This paper presents a novel data-driven crowd simulation method that can mimic the observed traffic of pedestrians in a given environment. Given a set of observed trajectories, we use a recent form of neural networks, Generative Adversarial Networks (GANs), to learn the properties of this set and generate new trajectories with similar properties. We define a way for simulated pedestrians (agents) to follow such a trajectory while handling local collision avoidance. As such, the system can generate a crowd that behaves similarly to observations, while still enabling real-time interactions between agents.
Via experiments with several real-world data sets, we show that our simulated trajectories preserve the statistical properties of their input. Our method can simulate crowds in real time that resemble existing crowds, while also allowing the insertion of extra agents, the combination with other simulation methods, and user interaction.
- Title: Data-Driven Crowd Simulation with Generative Adversarial Networks
- Authors: Javad Amirian, Wouter Van Toll, Jean-Bernard Hayet, Julien Pettré
- Download the full article: https://hal.inria.fr/hal-02134282/