InterMimicGen: Scaling Humanoid Loco-Manipulation through Self-Evolving Motion Imitation

2026-10-05

AI Summary

This study addresses the challenge that human motion data are sparse and heterogeneous, rendering them difficult to directly apply to humanoid mobile manipulation. To overcome this, we propose a self-evolving motion imitation framework that translates human demonstrations into robot commands via cross-embodiment contact-preserving retargeting, trains a generalizable whole-body tracking policy in physics simulation, and constructs a closed-loop data flywheel for the continuous augmentation and expansion of high-quality motion data. This work pioneers a unified pipeline from sparse human demonstrations to a scalable robot motion library, enabling a single policy to achieve broad action tracking. Successfully transferred to real-world robots, our approach significantly enhances both motion diversity and task execution success rates.

Abstract

Captured human-object interactions provide rich supervision for humanoid loco-manipulation, but they are sparse, heterogeneous, and not directly executable by robots. We introduce InterMimicGen, a self-evolving motion-imitation framework in which robot motion data and a tracking policy improve each other. First, we consolidate motion-captured human-object interaction datasets and retarget them into humanoid robot references while preserving whole-body coordination and dexterous hand-object relationships. This produces a large and diverse humanoid robot reference collection for dexterous whole-body loco-manipulation. Second, we train a physics-based generalist tracker that executes these references in simulation on a humanoid with dexterous hands, covering a scale and diversity beyond prior humanoid tracking systems for loco-manipulation. Third, we close a data flywheel: each round makes small, task-preserving changes to where an interaction takes place and how the body performs it, fine-tunes the tracker on them, and keeps only the variants whose simulated execution completes the task, which seed the next round. With more iterations, these small edits compound into broader coverage around the sparse original demonstrations while preserving task semantics and motion quality. Experiments show contact-preserving retargeting across robot configurations, broad tracking with a single generalist policy, executable motions that keep growing over augmentation rounds, and transfer to real robots. InterMimicGen provides a unified path from heterogeneous human demonstrations to a continually expanding motion resource for humanoid robot learning.

Problem

• Humanoid Loco-Manipulation
• Motion Imitation
• Human-Object Interaction
• Dexterous Manipulation

Innovation

• Humanoid Loco-Manipulation
• Self-Evolving Motion Imitation
• Data Flywheel
• Motion Retargeting
• Generalist Tracker