mimic-one: a Scalable Model Recipe for General Purpose Robot Dexterity

TL;DR

This paper introduces a diffusion-based control model for a highly dexterous humanoid robotic hand, achieving high success rates in complex real-world manipulation tasks through sample-efficient learning and emergent self-correcting behaviors.

Contribution

It presents a scalable model recipe combining hardware design, data collection, and generative control for dexterous robot manipulation, advancing practical real-world deployment.

Findings

Up to 93.3% success rate in real-world tasks

33.3% performance boost from self-correcting behaviors

Identifies scaling trends in policy performance

Abstract

We present a diffusion-based model recipe for real-world control of a highly dexterous humanoid robotic hand, designed for sample-efficient learning and smooth fine-motor action inference. Our system features a newly designed 16-DoF tendon-driven hand, equipped with wide angle wrist cameras and mounted on a Franka Emika Panda arm. We develop a versatile teleoperation pipeline and data collection protocol using both glove-based and VR interfaces, enabling high-quality data collection across diverse tasks such as pick and place, item sorting and assembly insertion. Leveraging high-frequency generative control, we train end-to-end policies from raw sensory inputs, enabling smooth, self-correcting motions in complex manipulation scenarios. Real-world evaluations demonstrate up to 93.3% out of distribution success rates, with up to a +33.3% performance boost due to emergent self-correcting behaviors, while also revealing scaling trends in policy performance. Our results advance the state-of-the-art in dexterous robotic manipulation through a fully integrated, practical approach to hardware, learning, and real-world deployment.