Soft and Compliant Contact-Rich Hair Manipulation and Care

Uksang Yoo1, Nathaniel Dennler2, Eliot Xing1, Maja Matarić2, Stefanos Nikolaidis2, Jeffrey Ichnowski1, Jean Oh1
IEEE/ACM International Conference on Human-Robot Interaction (HRI), 2025
1Carnegie Mellon University

2University of Southern California

Best Paper Finalist

Overview

MOE-Hair: A soft robotic system for safe and effective hair care

Abstract

Hair care robots have the potential to alleviate labor shortages in elderly care and enable those with limited mobility to express their identities through hair styling. We present MOE-Hair, a system that incorporates a compliant soft robotic manipulator and visual method for applied force estimation to perform three hair-care skills: head patting, finger combing, and hair grasping. MOE-Hair underscores two advantages of soft robotic manipulators in hair-care applications: safety through mechanical compliance and sensing force through observing deformation.

Key Findings

Safety Through Compliance

MOE soft robot hand applies 74% less force than rigid grippers while grasping comparable amounts of hair, demonstrating that soft robots can provide more comfortable care without sacrificing effectiveness.

Visual Force Estimation

Our novel force estimation module combines visual deformation data with tendon tensions to precisely track applied forces, reducing sensing errors by up to 60% compared to actuator-only approaches.

User Preference

Our study with 12 participants showed statistically significant preference for MOE with force estimation across all tasks. Participants commented the system felt "really similar to human fingers" and provided a sensation "like a head massage."

MOE: Multi-finger Omnidirectional End-effector

We introduce a dexterous tendon-driven soft robot manipulator that we call Multi-finger Omnidirectional End-effector (MOE) for hair-care applications.

Design Features:

  • Soft Fingers: Molded from low-hardness silicone (Ecoflex 00-30) for gentle contact
  • Tendon-Driven System: Two servomotors activate each finger through four embedded tendons
  • Wrist-Mounted Camera: RGBD camera provides egocentric view for force estimation
  • Human-Like Form Factor: 105mm length, 17mm diameter fingers designed for intuitive interaction

The design addresses a key challenge in previous works where human subjects tend to perceive rigid robots as being "rough" during hair care tasks. MOE's compliance makes it safer in unstructured environments and more robust in contact-rich manipulation tasks.

Hair Care Skills

Head Patting

MOE approaches the user's head from either the top or side to pat it, providing gentle contact with consistent force. The compliance of the fingers ensures safety during this interaction.

This task demonstrates MOE's ability to make initial contact with the head in a controlled, comfortable manner.

Finger Combing

MOE follows a user-defined trajectory across the user's head for finger combing, maintaining consistent contact with the scalp even as it moves along a path.

This task showcases the system's ability to maintain proper contact force during continuous movement, adapting to the contours of the head.

Hair Grasping

MOE approaches the user's head from either the top or side and grasps hair from near the scalp. The soft fingers allow for effective grasping while applying minimal force to the head.

In our tests, MOE applied 74% less force than rigid grippers while still grasping a comparable amount of hair.

Force Estimation Module

To enable MOE to maintain contact with the head during tasks and perform tasks effectively and comfortably for users, we developed methods for predicting MOE contact forces using:

  1. Visual Deformation: Using the wrist-mounted egocentric RGBD camera to capture depth images of MOE as it deforms during contact
  2. Tendon Tension: Observing the actuator current load which correlates to tension on the tendons

Our evaluation shows that the combined approach reduces sensing errors by up to 60.1% compared to actuator current load-only and 25.4% compared to depth image-only baselines.

User Study

We conducted a user study with 12 participants with diverse hair types to evaluate MOE-Hair across three key dimensions:

Task Effectiveness

Participants rated MOE-Hair as significantly more effective than the vision-only baseline across all three tasks.

User Comfort

Participants reported higher comfort levels, particularly in the more involved hair-grasping task.

Appropriate Use of Force

Significant preference for MOE-Hair's force application, especially during finger combing.

Participant Feedback:

"It felt really similar to human fingers... I kind of forgot it was a robot arm for a minute."

"It was like a head massage."

"I felt really comfortable."

Future Work & Extensions

We're extending MOE to various dexterity projects including:

This work demonstrates the unique advantages of soft robots in contact-rich hair-care tasks, while highlighting the importance of precise force control despite the inherent compliance of the system.

Citation

@inproceedings{yoo2025moehair,
  title={Soft and Compliant Contact-Rich Hair Manipulation and Care},
  author={Yoo, Uksang and Dennler, Nathaniel and Xing, Eliot and Mataric, Maja and Nikolaidis, Stefanos and Ichnowski, Jeffrey and Oh, Jean},
  booktitle={Proceedings of the 2025 ACM/IEEE International Conference on Human-Robot Interaction},
  year={2025}
}

Acknowledgements

This work is supported by the NSF GRFP (Grant No. DGE2140739) and MOTIE, Korea (Grant No. 20018112).