Summary: A new study from the University of Utah demonstrates that combining artificial intelligence with advanced fingertip proximity and pressure sensors enables a commercial bionic hand to grasp objects in a natural, intuitive way. An artificial neural network trained on common grasping postures lets each finger independently “see” an object and move to the correct position, improving grip security and precision while reducing cognitive effort for amputees.
Study participants were able to perform everyday tasks—lifting cups, picking up small items and other fine-motor activities—with far less mental strain and without extensive training. The shared-control design balances user intent and machine assistance, producing effortless, lifelike use of the prosthetic hand.
Key Facts
- Natural control: AI-driven fingertips use proximity and pressure sensors to enable stable, intuitive grasps.
- Reduced cognitive load: Users performed tasks with less mental effort and greater precision.
- Shared autonomy: A blended control strategy preserves user agency while augmenting dexterity.
Source: University of Utah
Our hands know how to grasp without thinking about each finger.
That effortless coordination is lost for many people who use prosthetic arms and hands. Even advanced robotic prostheses often force users to consciously command each motion, which adds cognitive burden and can make routine actions tiring or unreliable.
Researchers at the University of Utah addressed this challenge by outfitting a commercial TASKA Prosthetics hand with custom fingertips that combine sensitive pressure sensors and optical proximity sensors. These sensors detect minute contacts—small enough to feel the landing of a virtually weightless cotton ball—providing a high-resolution sense of the space immediately in front of each fingertip.
An artificial neural network was trained on proximity and pressure patterns associated with effective grasps. The model continuously positions each finger at the precise distance needed to form a secure, stable grip. Because every digit has its own sensor and operates in parallel, the prosthesis can adapt to different object shapes and sizes without the user having to plan each finger explicitly.
To preserve user agency, the team implemented a bioinspired shared-control framework that dynamically blends the user’s intent with the AI’s autonomous adjustments. The goal is not to override the wearer but to augment their natural control—improving precision and making tasks easier rather than forcing a user to fight the device for control.
“Nearly half of prosthesis users abandon their device, often citing cumbersome controls and mental fatigue,” said Marshall Trout, a postdoctoral researcher at the Utah NeuroRobotics Lab. “Here the machine augments the user’s natural control so tasks can be completed without extra thinking.”
The research team tested the system with four transradial amputee participants (amputations between the elbow and wrist) and with intact participants. Without extensive training, users performed standardized assessments and everyday tasks—such as drinking from a plastic cup or picking up small objects—with improved grip security and precision. Simple actions that previously required constant attention became more intuitive.
“By adding intelligence to the prosthesis, we offloaded the fine timing and positioning of fingers to the device itself,” said Jacob A. George, lead investigator and Solzbacher-Chen Endowed Professor in the Department of Electrical & Computer Engineering and a faculty member in Physical Medicine and Rehabilitation. “The result is more dexterous, more intuitive control that restores the simplicity of ordinary tasks.”
This work is part of the Utah NeuroRobotics Lab’s broader effort to enhance quality of life for amputees. The team is also exploring implanted neural interfaces that could let people control prostheses via intended movement and receive tactile sensations back from the device. Future plans aim to combine implanted neural control with the advanced fingertip sensors and AI-driven autonomy so the prosthesis can integrate seamlessly with thought-based commands and restore rich tactile function.
The study, led by Jacob A. George and Marshall Trout, was published in Nature Communications under the title “Shared human-machine control of an intelligent bionic hand improves grasping and decreases cognitive burden for transradial amputees.” Coauthors include members of the NeuroRobotics Lab—Fredi Mino, Connor Olsen and Taylor Hansen—along with Masaru Teramoto, David Warren and Jacob Segil.
Funding: Supported by the National Institutes of Health and the National Science Foundation.
Key Questions Answered:
A: The system combines fingertip proximity and pressure sensing with a trained neural network so each finger can automatically move into the position needed for a stable, natural grasp.
A: No. A shared-control framework dynamically blends human intent and machine assistance to prevent conflict and preserve user agency.
A: Conventional prostheses often demand high cognitive effort. This AI-driven approach restores low-effort, intuitive grasping that closely resembles natural hand function.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- The journal paper was reviewed in full.
- Additional context was supplied by the publication’s staff.
About this neurotech and robotics research news
Author: Evan Lerner
Source: University of Utah
Contact: Evan Lerner – University of Utah
Image: The image is credited to Neuroscience News
Original Research: Open access. Title: “Shared human-machine control of an intelligent bionic hand improves grasping and decreases cognitive burden for transradial amputees” by Jacob A. George et al., published in Nature Communications.
Abstract
Shared human-machine control of an intelligent bionic hand improves grasping and decreases cognitive burden for transradial amputees
Bionic hands recreate many movements of the human hand, but intuitive control remains limited. Human manual dexterity relies on sensory feedback and subconscious models that anticipate hand-object interactions. In this study, proximity and pressure sensors were integrated into a commercial prosthesis to enable autonomous finger positioning, while users controlled overall grasp via surface electromyography.
A bioinspired, dynamically weighted scheme merged machine decisions and user intent. This shared control produced greater grip security and precision and reduced cognitive burden. Demonstrations with both intact and amputee participants show the modified prosthesis performing real-world tasks across multiple grip patterns. Granting some autonomy to bionic hands offers a practical route to more dexterous, intuitive prosthetic function.