Soft and Compliant Contact-Rich Hair Manipulation and Care

TL;DR

This work introduces MOE-Hair, a soft tendon-driven end-effector for contact-rich hair care tasks, enabled by a wrist-mounted RGBD camera and a force-estimation module that fuses visual deformation with actuator currents. The MOE design achieves substantially lower head contact forces than rigid grippers while maintaining hair-grasping efficacy, and the force-estimation approach improves accuracy over baselines that rely solely on depth or actuator data. A user study with 12 participants shows MOE-Hair is preferred for effectiveness, comfort, and appropriate force across head patting, finger combing, and hair grasping. The results advocate for soft robots in assistive, contact-rich tasks and point to future work on trajectory personalization and user-specific force thresholds.

Abstract

Hair care robots can help address labor shortages in elderly care while enabling those with limited mobility to maintain their hair-related identity. We present MOE-Hair, a soft robot system that performs three hair-care tasks: head patting, finger combing, and hair grasping. The system features a tendon-driven soft robot end-effector (MOE) with a wrist-mounted RGBD camera, leveraging both mechanical compliance for safety and visual force sensing through deformation. In testing with a force-sensorized mannequin head, MOE achieved comparable hair-grasping effectiveness while applying significantly less force than rigid grippers. Our novel force estimation method combines visual deformation data and tendon tensions from actuators to infer applied forces, reducing sensing errors by up to 60.1% and 20.3% compared to actuator current load-only and depth image-only baselines, respectively. A user study with 12 participants demonstrated statistically significant preferences for MOE-Hair over a baseline system in terms of comfort, effectiveness, and appropriate force application. These results demonstrate 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.

Figures (15)

• Figure 1: Design of proposed MOE end-effector. Left: exploded view and assembly of MOE. Right: fully assembled MOE.
• Figure 2: MOE in varying poses. MOE deforms easily with contact by inherent passive compliance of the fingers' material. As the fingers deform, they apply more tension on the tendons, which our proposed method uses in conjunction with depth image to estimate applied forces. The tendons can be actuated in various ways to achieve compliant dexterity of the fingers.
• Figure 3: Setup for training data collection. We use a controlled setup with a force sensor and allow MOE to randomly sample various contact conditions, capturing training observations and forces in a self-labeling manner to scale up data collection.
• Figure 4: Proposed method for force estimation. The proposed applied force estimation module uses segmented depth image and actuator current loads as input to predict the force vector.
• Figure 5: Forces during hair grasping. We carried out the experiments at three different depths into the hair. The depth measurements were from the point where the end-effector just made contact with the hair to account for different lengths. MOE exerts measurably less force and torque on the head.