From Flow to One Step: Real-Time Multi-Modal Trajectory Policies via Implicit Maximum Likelihood Estimation-based Distribution Distillation

TL;DR

This paper introduces a fast, single-step policy distillation method for multi-modal robotic control, combining implicit maximum likelihood estimation with a bi-directional Chamfer distance to preserve diverse behaviors and enable real-time, high-frequency decision-making.

Contribution

It presents a novel IMLE-based distillation framework with a set-level loss and a unified perception encoder, enabling real-time, multi-modal control with preserved diversity.

Findings

Achieves real-time control with high-frequency re-planning.

Preserves multi-modal distribution in a single forward pass.

Improves robustness under dynamic disturbances.

Abstract

Generative policies based on diffusion and flow matching achieve strong performance in robotic manipulation by modeling multi-modal human demonstrations. However, their reliance on iterative Ordinary Differential Equation (ODE) integration introduces substantial latency, limiting high-frequency closed-loop control. Recent single-step acceleration methods alleviate this overhead but often exhibit distributional collapse, producing averaged trajectories that fail to execute coherent manipulation strategies. We propose a framework that distills a Conditional Flow Matching (CFM) expert into a fast single-step student via Implicit Maximum Likelihood Estimation (IMLE). A bi-directional Chamfer distance provides a set-level objective that promotes both mode coverage and fidelity, enabling preservation of the teacher multi-modal action distribution in a single forward pass. A unified perception encoder further integrates multi-view RGB, depth, point clouds, and proprioception into a geometry-aware representation. The resulting high-frequency control supports real-time receding-horizon re-planning and improved robustness under dynamic disturbances.