Upper limb exoskeleton research for overhead task assistance

1. Funding Agency

Samsung Heavy Industry

2. Motivation

Repetitive overhead tasks in shipbuilding expose workers to high physical loads, particularly on the shoulder, leading to an increased risk of musculoskeletal disorders (MSDs) and fatigue-related performance decline. Conventional passive exoskeletons often lack adaptability and require manual adjustment. To reduce muscular effort and improve endurance during heavy tool handling, there is a growing demand for lightweight, energy-efficient wearable robotic solutions that can dynamically support upper-limb movements with minimal user burden and high usability in industrial environments.

3. Objectives

This project aims to design and develop a lightweight, semi-passive shoulder exoskeleton using the Twisted Rubber Stiffness Tuner (TRUST). TRUST enables stiffness modulation through twist angle control without clutches or high-torque motors. The exoskeleton targets upper-limb support for overhead tasks by converting TRUST’s restoring force into joint torque via a gravity compensation mechanism. The system seeks to minimize energy consumption while ensuring sufficient torque output (up to 8.11 Nm), and maintain a compact hardware form factor (2.35 kg) suitable for prolonged industrial use.

4. Results

The TRUST-based shoulder exoskeleton delivers 3.01–8.11 Nm of shoulder joint torque while weighing only 2.35 kg. Its torque-to-weight ratio (TWR) and adjustable torque range (TWR margin) significantly outperform other semi-passive systems. A low-energy controller minimizes motor activity by exploiting gear friction to hold twist angles, resulting in low energy consumption (validated through TER and TTR metrics). The system dynamically adapts to varying tool weights and postures using high-level controllers, demonstrating both mechanical efficiency and ergonomic suitability for extended use in industrial tasks.

5. Related Publications

• H.-S. Seong, S. Farag, S.-W. Lee, I. Gaponov, and J.-H. Ryu,“Wearable robot control method based on vision-language models,”in First Workshop on Vision-Language Models for Navigationand Manipulation at ICRA 2024.
• Twisted Rubber Stiffness Tuner (고무 꼬임 강성 조절 장치), PCT patent application in progress

6. Contact

• Hyeonseok Seong