Custom 3D-Printed Sleeve Decodes 19 Gestures for Prosthetic Hand Control

Current prosthetic hands often fail to adapt to the unique anatomy of each amputee, leading to difficulties in achieving natural and intuitive control. This mismatch can result in a frustrating user experience, sometimes causing individuals to abandon the devices. While advancements have been made in interpreting muscle signals, their inherent instability and translation challenges remain a core problem.

To address this, researchers are focusing on personalized systems rather than standardized devices. The process begins with a 3D scan of a user's residual limb to create a custom, 3D-printed wearable sleeve. This sleeve integrates soft, flexible magnetic sensors that rest against the skin, capturing subtle muscle shape and pressure changes during intended hand and wrist movements. This allows the system to interpret user intent in real time.

The sensor array configuration is tailored to individual limb size and anatomy, with either 18 or 24 modules. This is paired with an individualized artificial intelligence model that learns each person's unique muscle patterns, moving away from generalized datasets. In tests involving 10 participants, including three upper-limb amputees, the system successfully classified 19 hand and wrist gestures in real time, translating intent into control of a dexterous robotic hand.

Durability testing included over 7,500 robotic force cycles, demonstrating a stable and strong relationship between applied force and sensor output without performance loss. Even after extensive use, the sensor signals remained clear and stable, showing minimal drift or degradation. These findings underscore the importance of matching sensor placement and quantity to the individual's anatomy and remaining muscle function for optimal prosthetic performance.