A Closed-Loop CPR Training Glove with Integrated Tactile Sensing and Haptic Feedback

A research team from multiple institutions, led by Jaeyoung Moon, has developed a novel closed-loop CPR training glove designed to enable effective, self-directed practice. The glove integrates a high-resolution tactile sensing array across the palm and dorsum to measure distributed pressures in real-time. This data feeds into lightweight statistical models that achieve over 92% accuracy for force estimation and hand pose classification, with inference times under a millisecond. The system continuously models key CPR performance metrics—compression rate, force, and hand position—creating a closed-loop pipeline for immediate correction.

Based on these real-time estimations, the glove delivers haptic feedback through built-in vibrotactile actuators to guide the user toward proper CPR technique. An 8-person user study showed this tactile guidance reduced visual distraction compared to traditional audio-visual cues, though the team notes that simplified vibration patterns were necessary for reliable perception during the dynamic physical load of compressions. The tactile sensor itself was rigorously quantified, showing a sensitivity of ~0.85 over a 0-600 N range, with an 11.05% signal drift over 300 cycles. The paper, accepted for ICRA 2026, highlights the system's feasibility and provides design insights for future wearable haptic training tools.