Eric Wai Ming Lee

The paper proposes CPGAN, a novel Generative Adversarial Network (GAN) that incorporates a collision-penalizing loss function to significantly improve the simulation of collision avoidance in dense, bidirectional crowd movements.