LASIL: Learner-Aware Supervised Imitation Learning For Long-term Microscopic Traffic Simulation

Addressing the covariate shift problem in multi-agent imitation learning for realistic traffic simulation.