From Flow to One Step

Real-Time Multi-Modal Trajectory Policies via Implicit Maximum

Likelihood Estimation-based Distribution Distillation