DYMO-Hair: Generalizable Volumetric Dynamics Modeling for Robot Hair Manipulation
Chengyang Zhao, Uksang Yoo, Arkadeep Narayan Chaudhury, Giljoo Nam, Jonathan Francis, Jeffrey Ichnowski, Jean Oh
AI summary
Problem
Routine hair care is inaccessible for individuals with limited mobility, while existing robotic systems fail to generalize across diverse hairstyles due to hair's complex, fine-grained physical dynamics and lack of scalable data.
Approach
The system represents hair as a high-resolution 3D volumetric grid with orientation fields, pre-trains a compact latent space for diverse hairstyles, and learns dynamics as action-conditioned state editing, all powered by a novel strand-level physics simulator for synthetic data generation.
Key results
- First 3D generalizable volumetric dynamics model for hair combing
- Novel GPU-accelerated strand-level PBD simulator for synthetic data generation
- 22% lower geometric error and 42% higher success rate in simulation vs. baselines
- Zero-shot real-world transfer to physical wigs with consistent success on unseen styles
Why it matters
Provides a foundational framework for accessible, autonomous robot hair care that generalizes across diverse real-world hairstyles, advancing deformable object manipulation for assistive robotics.
Abstract
Hair care is an essential daily activity, yet it remains inaccessible to individuals with limited mobility and challenging for autonomous robot systems due to the fine- grained physical structure and complex dynamics of hair. In this work, we present DYMO-HAIR, a model-based robot hair care system. We introduce a novel dynamics learning paradigm that is suited for volumetric quantities such as hair, relying on an action-conditioned latent state editing mechanism, coupled with a compact 3D latent space of diverse hairstyles to improve generalizability. This latent space is pre-trained at scale using a novel hair physics simulator, enabling generalization across previously unseen hairstyles. Using the dynamics model with a Model Predictive Path Integral (MPPI) planner, DYMO-HAIR is able to perform visual goal-conditioned hair styling. Experi- ments in simulation demonstrate that DYMO-Hair’s dynamics model outperforms baselines on capturing local deformation for diverse, unseen hairstyles. DYMO-Hair further outperforms baselines in closed-loop hair styling tasks on unseen hairstyles, with an average of 22% lower final geometric error and 42% higher success rate than the state-of-the-art system. Real-world experiments exhibit zero-shot transferability of our system to wigs, achieving consistent success on challenging unseen hairstyles where the state-of-the-art system fails. Together, these results introduce a foundation for model-based robot hair care, advancing toward more generalizable, flexible, and accessible robot hair styling in unconstrained physical environments. More details can be found at: https://dymohair.github.io.