Adaptive Multi-Sensory Augmented Reality Interface for Operator Training in Smart Manufacturing Cells Using Real-Time Biosignal Feedback and Edge-AI
Abstract
The growing complexity of modern manufacturing demands that operators rapidly acquire and retain advanced operational skills for complex machinery. Conventional training tools, such as static augmented reality (AR) interfaces, fail to address realtime challenges including operator stress, cognitive load, and fluctuating engagement levels. These limitations hinder learning effectiveness, prolong onboarding, and increase the likelihood of errors or safety incidents. To overcome these issues, this study presents an Adaptive Multi-Sensory Augmented Reality Interface (AMARTS) that uniquely integrates real-time biosignal feedback with Edge-AI-based adaptive training for smart manufacturing environments. The proposed system employs nine non-invasive biosensors embedded in AR headsets and smart gloves to continuously monitor physiological signals such as heart rate variability and skin conductance-key indicators of stress and fatigue. A lightweight Edge-AI model processes these biosignals locally with minimal latency to classify the operator's state while preserving data privacy. Based on the realtime analysis, the AR interface dynamically adjusts visual, auditory, and haptic feedback to align with the operator's cognitive and emotional conditions, ensuring sustained engagement and optimal skill acquisition. Pilot trials conducted in simulated manufacturing conditions demonstrate that AMARTS achieves superior performance with <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 5. 8 \%}$</tex> sensitivity, <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 1. 2 \%}$</tex> specificity, and 96.6% overall accuracy, while reducing training errors and improving operator confidence and safety. The integration of biosignal-driven adaptation and edgeintelligent learning establishes a novel, human-centric framework for operator training aligned with the evolving demands of Industry <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{4. 0}$</tex>.