An intelligent taekwondo coaching system based on augmented reality technology with real-time feedback mechanisms

TL;DR

This paper introduces an augmented reality-based taekwondo coaching system that provides real-time feedback to improve training efficiency and technique accuracy.

Contribution

The novel integration of AR, deep learning, and biomechanical analysis for real-time, objective taekwondo coaching.

Findings

The system achieves over 95% accuracy in recognizing nine taekwondo techniques with less than 25ms latency.

47 practitioners showed improved learning efficiency and technique standardization compared to traditional methods.

Users reported high satisfaction (8.5/10) with the system's usability and feedback clarity.

Abstract

Traditional taekwondo training methods face limitations in providing objective, real-time feedback for technique improvement, relying primarily on subjective instructor observations that may lack precision and consistency. This research presents an innovative intelligent taekwondo coaching framework that integrates augmented reality technology with advanced motion analysis algorithms to deliver comprehensive, real-time training feedback. The system employs a modular architecture incorporating multi-modal sensor data acquisition, deep learning-based pose estimation, biomechanical analysis, and immersive AR visualization to create an interactive training environment. The motion recognition module utilizes convolutional neural networks specifically adapted for taekwondo techniques, achieving recognition accuracies exceeding 95% across nine fundamental technique categories with processing latencies below 25 milliseconds. The comprehensive evaluation framework assesses movement quality across eight dimensions including geometric accuracy, temporal coordination, and force generation, providing personalized feedback through AR overlays. Experimental validation with 47 practitioners across novice, intermediate, and advanced skill levels demonstrates significant improvements in learning efficiency and technique standardization compared to conventional training methods. User experience evaluation reveals high satisfaction ratings (average 8.5/10) across interface usability, feedback clarity, and learning effectiveness. The system successfully addresses critical gaps in martial arts instruction by democratizing access to high-quality technical guidance while maintaining the engagement and motivation essential for skill development. This research establishes important foundations for next-generation sports training technologies and provides valuable insights for developing similar intelligent coaching systems across diverse athletic disciplines.

Figures