Researchers at Florida Atlantic University (FAU) have developed a new approach to prosthetic hand control that combines 3D scanning, 3D printing, soft magnetic sensors and artificial intelligence to create a more personalised interface between the user and a robotic hand.
The work, reported by Hackster.io, addresses one of the major limitations in advanced upper-limb prosthetics: modern prosthetic hands can be mechanically capable, but many users still struggle to control them naturally and consistently. The challenge is that biological signals vary widely between users and can also change over time because of fatigue, posture, residual limb anatomy, sensor placement and other factors. (Hackster.io)
FAU’s system starts with a 3D scan of the user’s residual limb. That scan is used to design and print a custom wearable sleeve that fits the individual user. The sleeve contains either 18 or 24 compliant magnetic force myography sensor modules, depending on the person’s anatomy. Instead of relying on electrical muscle signals, as in conventional electromyography-based systems, the sensors detect subtle changes in muscle shape and pressure as the user attempts different hand and wrist movements. (Hackster.io)
Each sensor module includes a soft silicone structure with an embedded neodymium magnet positioned above a Hall effect sensor. As the muscles move, the magnetic field changes, allowing the system to capture movement intent. Hackster notes that this magnetic force myography approach may avoid some of the practical limitations associated with EMG, including issues linked to sweat and changing skin conductivity. (Hackster.io)
The sensor data is paired with an individualised AI model trained around the user’s own movement patterns. FAU describes the goal as moving away from one-size-fits-all prosthetic control and towards systems that adapt to the person. In testing, the system was evaluated with 10 participants, including three upper-limb amputees, and classified 19 hand and wrist gestures in real time. FAU reported stable real-time control of multiple gestures under repeated use, while Hackster reported a mean accuracy of 93.64% across participants. (Florida Atlantic University, Hackster.io)
The research has been published in IEEE Transactions on Neural Systems and Rehabilitation Engineering, and FAU says the team has also released the resulting dataset as an open resource for the prosthetics research community. (Florida Atlantic University)
For prosthetists and rehabilitation teams, the most important message is not simply that the system uses AI or 3D printing. It is that prosthetic control may increasingly depend on individualised hardware and individualised software working together. The best sensor layout was not the same for every participant, suggesting that future upper-limb systems may need more personalised clinical fitting processes rather than standardised sensor placement.
This has practical implications for the prosthetics and orthotics sector across the Middle East, Africa, Central Asia and South Asia. As advanced upper-limb technologies become more available, clinicians may need to build skills in digital capture, wearable sensor integration, data-informed fitting and long-term device tuning. Manufacturers and service providers may also need to consider how custom 3D-printed interfaces can be produced reliably and affordably in regional rehabilitation settings.
The FAU work also highlights a wider shift in prosthetic development: the socket or wearable interface is no longer only a mechanical connection between body and device. It can also become a sensing platform, capturing information from the residual limb and translating it into more intuitive movement. If this approach can be made robust, affordable and clinically practical, it could support better user confidence, improved control and potentially lower device abandonment in upper-limb prosthetics.
While further clinical testing and real-world validation will be needed, the project offers a clear direction for the future: prosthetic hands that do not only move well, but also learn how each person intends to move.
- Hackster.io: The Future of Prosthetics Is 3D-Printed and AI-Powered
- Florida Atlantic University: FAU Engineers Prosthetic Hand That Learns, Adapts to Each User
- IEEE Xplore: Compliant Magnetic Sensor Arrays Enable Real-Time Force Myogram Pattern Recognition for Dexterous Hand Control by Amputees
- PubMed: Compliant Magnetic Sensor Arrays Enable Real-Time Force Myogram Pattern Recognition
- Florida Atlantic University College of Engineering and Computer Science
- WHO: Assistive technology
- ISPO: International Society for Prosthetics and Orthotics