Project/ Kindex — Hongyu Yue's works

About

KindexA Smart Garment for Muscle Compensation Recognition

Overview
Compensation is the unconscious use of other muscle groups to functionally replace fatigued or injured muscles, thereby complicating rehabilitation and injury prevention.
KinDex is a fully textile-integrated wearable system combined with neural networks to detect compensatory movements and provide real-time feedback, supporting effective home-based rehabilitation training.

ContributionIndividual (Capstone Project)

DurationSep 2025 – May 2026

InstructorQi Wang (Head of Center for Digital Innovation, Tongji University)
Nianchong Qu (PhD at Center for Digital Innovation, Tongji University)

KeywordsSmart Textiles, Rehabilitation, Deep Learning, Wearable Device

StatusAccepted by Ubicomp/ISWC 2026 Design Exhibition

Please view in a full-screen browser window for the best experience.

ProblemsMuscle compensation often occurs after injury or overuse. When a weakened muscle can no longer perform a movement properly, surrounding muscles take over to compensate. This adaptation may create secondary injuries, worsen the original problem, and often goes unnoticed. For example, during shoulder abduction, users may unconsciously compensate through shoulder shrugging or internal rotation. Traditional rehabilitation relies on frequent clinical visits, but subtle compensation patterns are difficult to identify and correct outside the clinic. My research explores a home-based approach for accurate compensation recognition and correction.

Gaps in Existing Research  Existing approaches either fail to capture subtle muscle-level changes or require complex laboratory environments, limiting their use in daily rehabilitation.
An ideal solution would be a wearable system integrated into everyday clothing that can continuously monitor movement and detect compensation patterns in real-world settings.

Research DirectionA fully textile garment that supports long-term wear, enabling real-time muscle compensation detection and feedback-based correction during shoulder rehabilitation.

Core ApproachThe core idea of this system is to integrate multiple textile stretch sensors onto a wearable garment. As the user performs rehabilitation movements, the textile stretch sensors stretch with body motion. The silver-coated conductive fibers inside the sensors change resistance when stretched, allowing the system to capture subtle changes in muscle and body movement.

By combining multi-location stretch signals with a Bidirectional LSTM network, the system learns to recognize different compensation patterns. It can identify whether the user is compensating through shoulder elevation, internal rotation, or posterior movement, and provide real-time feedback through audio or UI guidance to help correct the movement.

Garment DesignTo obtain stable and reliable sensing data, the stretch sensors and conductive fibers must be securely integrated into the fabric while remaining free from unwanted stretching or mechanical interference. This imposes strict requirements on garment structural design. I therefore performed multiple iterations and optimizations across garment patterning and routing, sensor placement and fixation, and integration of the driver module.

Garment DesignTo

The system adopts textile stretch sensors, whose resistance changes with stretching to collect muscle motion data. Based on rehabilitation motion simulation and experimental testing, 14 sensors are placed in key areas with the highest information density and fixed with elastic stitching.

Sensors are connected to the driver board via 16 fiber wires. The wires are fixed inside the garment with elastic stitching, and the layout is optimized to minimize mechanical interference with sensor deformation.

Ongoing WorkRepeated donning and doffing, along with extended movement, cause relative displacement between the garment and skin. Additionally, the generic sensor layout still leaves room for improved information efficiency. My next phase will focus on systematically refining sensor structure, layout, and garment architecture to enhance signal reliability during prolonged use. 

Repeated donning and doffing, along with extended movement, cause relative displacement between the garment and skin. Additionally, the generic sensor layout still leaves room for improved information efficiency. My next phase will focus on systematically refining sensor structure, layout, and garment architecture to enhance signal reliability during prolonged use. 

All Works

Hongyu Yue 岳洪宇