MiVLA: Towards Generalizable Vision-Language-Action Model with Human-Robot Mutual Imitation Pre-training

TL;DR

MiVLA introduces a mutual imitation pre-training approach leveraging human and robot data to improve generalization in vision-language-action models for robotic control tasks.

Contribution

The paper presents MiVLA, a novel framework that uses behavioral similarity and kinematic alignment for mutual imitation, enhancing generalization across different embodiments and environments.

Findings

Outperforms state-of-the-art VLAs by 25% in simulation.

Achieves 14% improvement in real-world robot control.

Demonstrates strong generalization on multiple robots.

Abstract

While leveraging abundant human videos and simulated robot data poses a scalable solution to the scarcity of real-world robot data, the generalization capability of existing vision-language-action models (VLAs) remains limited by mismatches in camera views, visual appearance, and embodiment morphologies. To overcome this limitation, we propose MiVLA, a generalizable VLA empowered by human-robot mutual imitation pre-training, which leverages inherent behavioral similarity between human hands and robotic arms to build a foundation of strong behavioral priors for both human actions and robotic control. Specifically, our method utilizes kinematic rules with left/right hand coordinate systems for bidirectional alignment between human and robot action spaces. Given human or simulated robot demonstrations, MiVLA is trained to forecast behavior trajectories for one embodiment, and imitate behaviors for another one unseen in the demonstration. Based on this mutual imitation, it integrates the behavioral fidelity of real-world human data with the manipulative diversity of simulated robot data into a unified model, thereby enhancing the generalization capability for downstream tasks. Extensive experiments conducted on both simulation and real-world platforms with three robots (ARX, PiPer and LocoMan), demonstrate that MiVLA achieves strong improved generalization capability, outperforming state-of-the-art VLAs (e.g., $\boldsymbol{\pi}_{0}$, $\boldsymbol{\pi}_{0.5}$ and H-RDT) by 25% in simulation, and 14% in real-world robot control tasks.