<p>Hand impairments caused by stroke or neuromuscular disorders severely restrict daily activities, underscoring the need for effective, home-applicable rehabilitation systems. This study presents HandMaid, a wearable hand exoskeleton that restores motion through electromyography (EMG)-based intention detection and an active backdrive safety mechanism. The system integrates a 3D-printed rigid exoskeleton, linear actuators, and a neural network interface that is retrained before each use to compensate for variations in electrode placement, skin resistance, temperature, and humidity. This adaptive training procedure ensures reliable signal interpretation under different environmental and physiological conditions. Kinematic and experimental analyses demonstrate that the device covers approximately 81% of the anatomical range of motion (MCP <InlineEquation ID="IEq1"> <EquationSource Format="TEX">$\approx $</EquationSource> <EquationSource Format="MATHML"><math> <mo>≈</mo> </math></EquationSource> </InlineEquation> 50<InlineEquation ID="IEq2"> <EquationSource Format="TEX">$^{\circ }$</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mrow /> <mo>∘</mo> </mmultiscripts> </math></EquationSource> </InlineEquation>, PIP <InlineEquation ID="IEq3"> <EquationSource Format="TEX">$\approx $</EquationSource> <EquationSource Format="MATHML"><math> <mo>≈</mo> </math></EquationSource> </InlineEquation> 95<InlineEquation ID="IEq4"> <EquationSource Format="TEX">$^{\circ }$</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mrow /> <mo>∘</mo> </mmultiscripts> </math></EquationSource> </InlineEquation>, DIP <InlineEquation ID="IEq5"> <EquationSource Format="TEX">$\approx $</EquationSource> <EquationSource Format="MATHML"><math> <mo>≈</mo> </math></EquationSource> </InlineEquation> 90<InlineEquation ID="IEq6"> <EquationSource Format="TEX">$^{\circ }$</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mrow /> <mo>∘</mo> </mmultiscripts> </math></EquationSource> </InlineEquation>), while maintaining a mean classification accuracy of 98.1 ± 1.6% and an average response latency of <InlineEquation ID="IEq7"> <EquationSource Format="TEX">$\approx $</EquationSource> <EquationSource Format="MATHML"><math> <mo>≈</mo> </math></EquationSource> </InlineEquation>1.05 s. The threshold-based active backdrive logic provides immediate reversal in case of user resistance or discomfort, ensuring both safety and comfort during use. Weighing around 400 g and with a total production cost below $350, HandMaid offers a portable, low-cost, and anatomically consistent rehabilitation solution. The results confirm that the system enables user-safe, adaptive, and cost-effective robotic rehabilitation suitable for both clinical and home environments, while its retrainable EMG model ensures robustness against session-to-session variability.</p>

Handmaid: design and evaluation of a hand exoskeleton with active backdrive for neuromuscular rehabilitation

期刊论文
出版模式：

开放获取
发表日期：

2026年2月13日

查看全文

Hakkı Güçlü,
Adnan Cora

摘要

Hand impairments caused by stroke or neuromuscular disorders severely restrict daily activities, underscoring the need for effective, home-applicable rehabilitation systems. This study presents HandMaid, a wearable hand exoskeleton that restores motion through electromyography (EMG)-based intention detection and an active backdrive safety mechanism. The system integrates a 3D-printed rigid exoskeleton, linear actuators, and a neural network interface that is retrained before each use to compensate for variations in electrode placement, skin resistance, temperature, and humidity. This adaptive training procedure ensures reliable signal interpretation under different environmental and physiological conditions. Kinematic and experimental analyses demonstrate that the device covers approximately 81% of the anatomical range of motion (MCP $\approx $ $\approx$ 50 $^{\circ }$ $^{\circ}$ , PIP $\approx $ $\approx$ 95 $^{\circ }$ $^{\circ}$ , DIP $\approx $ $\approx$ 90 $^{\circ }$ $^{\circ}$ ), while maintaining a mean classification accuracy of 98.1 ± 1.6% and an average response latency of $\approx $ $\approx$ 1.05 s. The threshold-based active backdrive logic provides immediate reversal in case of user resistance or discomfort, ensuring both safety and comfort during use. Weighing around 400 g and with a total production cost below $350, HandMaid offers a portable, low-cost, and anatomically consistent rehabilitation solution. The results confirm that the system enables user-safe, adaptive, and cost-effective robotic rehabilitation suitable for both clinical and home environments, while its retrainable EMG model ensures robustness against session-to-session variability.