Particle.news
Download on the App Store

AI-Guided Bionic Hand From University of Utah Improves Grip in Peer-Reviewed Human Study

Custom fingertips with optical proximity sensing let an AI adjust each finger before contact for more natural, low-effort grasping.

Overview

  • The team modified a commercial TASKA hand with pressure sensors and optical proximity sensors that detect objects just before touch.
  • An artificial neural network trained on proximity data autonomously fine-tunes each digit’s position to form stable, precise grasps.
  • A bioinspired shared-control scheme balances user intent with AI assistance so the prosthesis augments control rather than competing with it.
  • Four transradial amputees showed better performance on standardized tasks and completed fine-motor activities with less mental effort, according to the Nature Communications paper.
  • Researchers plan to integrate implanted neural interfaces for thought-based control and restored touch, with broader clinical testing and commercialization not yet announced.