Augmented Reality Systems for Skill-Based Digital Training

Abstract

Augmented reality (AR) overlays digital information -- step-by-step instructions, 3D component annotations, safety alerts, and performance feedback -- onto the real-world training environment, enabling hands-on skill development with real equipment while providing contextual digital guidance unavailable in traditional instruction. Unlike virtual reality's fully immersive simulated environment, AR preserves physical task engagement -- trainees use real tools, feel real materials, and develop genuine muscle memory -- while digital overlays reduce cognitive load from instruction lookup, prevent errors through real-time verification, and provide immediate corrective feedback. This paper proposes the AR Skill Training Evaluation Framework (ARSTEF), a systematic evaluation of five AR system architectures -- smart glasses (OST-AR), handheld tablet AR, projected AR, wearable wrist AR, and AI-enhanced spatial AR -- across six skill training domains: industrial maintenance, surgical technique, culinary training, electronic assembly, construction inspection, and military equipment operation. ARSTEF involves 2,880 trainees across 18 training programmes over 12 months, introducing the AR Training Effectiveness Score (ARTES) integrating skill acquisition rate, error reduction, training time savings, and system usability. Key results: AI-enhanced spatial AR achieves the highest ARTES (0.912), reducing training time by 38.4% and error rate by 52.4% versus instruction manual baseline; OST smart glasses achieve best hands-free usability (SUS = 84.2); projected AR achieves highest spatial precision for electronic assembly (0.02 mm placement accuracy). The framework provides AR system selection guidance across skill training domains.