Training with agency‐inspired feedback from an instrumented glove to improve functional grasp performance

Mingxiao Liu, Samuel Wilder, Sean Sanford, Soha Saleh, Noam Y. Harel, Raviraj Nataraj

Research output: Contribution to journalArticlepeer-review

9 Scopus citations

Abstract

Sensory feedback from wearables can be effective to learn better movement through en-hanced information and engagement. Facilitating greater user cognition during movement practice is critical to accelerate gains in motor function during rehabilitation following brain or spinal cord trauma. This preliminary study presents an approach using an instrumented glove to leverage sense of agency, or perception of control, to provide training feedback for functional grasp. Seventeen able‐bodied subjects underwent training and testing with a custom‐built sensor glove prototype from our laboratory. The glove utilizes onboard force and flex sensors to provide inputs to an artificial neural network that predicts achievement of “secure” grasp. Onboard visual and audio feedback was provided during training with progressively shorter time delay to induce greater agency by intentional binding, or perceived compression in time between an action (grasp) and sensory consequence (feedback). After training, subjects demonstrated a significant reduction (p < 0.05) in movement pathlength and completion time for a functional task involving grasp‐move‐place of a small object. Future work will include a model‐based algorithm to compute secure grasp, virtual reality immersion, and testing with clinical populations.

Original languageEnglish
Article number1173
Pages (from-to)1-15
Number of pages15
JournalSensors
Volume21
Issue number4
DOIs
StatePublished - 2 Feb 2021

Keywords

  • Artificial intelligence
  • Cognition
  • Hand
  • Hand strength
  • Rehabilitation
  • Sensory feedback

Fingerprint

Dive into the research topics of 'Training with agency‐inspired feedback from an instrumented glove to improve functional grasp performance'. Together they form a unique fingerprint.

Cite this