University of Utah engineers developed an AI-enabled prosthetic hand that uses fingerprint-like proximity and pressure sensors plus machine learning to grasp small objects more naturally. The sensors can detect very light touches—reportedly as sensitive as a cotton ball landing on a fingertip—and the AI positions fingers at the ideal distance for a secure hold. Researchers also tuned how control is shared between user and device to prevent conflict. The study was published in Nature Communications on Dec. 9, 2025.
AI Prosthetic Hand Learns to Grasp Like a Human, University of Utah Researchers Report
Dec 11
Lead author Marshall Trout, right, worked with four amputees to investigate how AI could be used to autonomously control an advanced prothesis. (Image credit: Utah NeuroRobotics Lab)
Engineers at the University of Utah have developed an AI-enabled prosthetic hand that combines sensitive fingertip sensors with machine learning to produce more natural, intuitive grasps of small objects.
How It Works
The research team outfitted an artificial hand with custom, fingerprint-like ridges embedded with proximity and pressure sensors designed to replicate a fine sense of touch. In laboratory tests, the sensors were sensitive enough to detect something as light as a cotton ball landing on the fingers.
An AI model was trained to recognize common grasping postures and to move the fingers to "the exact distance necessary to form a perfect grasp of the object." This enables the prosthesis to approach and close around objects in a manner that feels more natural and requires less conscious correction from the user.
Balancing Control
Researchers also tuned how control is shared between the user and the device, avoiding situations where the human and machine fight for command. "What we don’t want is the user fighting the machine for control. In contrast, here the machine improved the precision of the user while also making the tasks easier," said Marshall Trout, an engineering professor at the University of Utah.
“By adding some artificial intelligence, we were able to offload this aspect of grasping to the prosthesis itself,” said Jacob A. George, who co-led the study. “The end result is more intuitive and more dexterous control, which allows simple tasks to be simple again.”
The study was coauthored by members of the NeuroRobotics Lab and a professor at the University of Colorado Boulder. The research was published in Nature Communications on Dec. 9, 2025.
Why It Matters
Many prosthesis users abandon their devices because controls are unintuitive and mentally burdensome; the University of Utah's approach aims to reduce that cognitive load by letting the prosthetic handle routine aspects of grasping. If translated to clinical devices, this technology could improve daily function and user satisfaction for people with upper-limb loss.
