AI‐Driven Multi‐Modal Information Fusion System for Real Applications in Physical Training

Abstract

This paper introduces an AI‐driven multi‐modal information fusion system tailored for real‐world physical training applications, integrating state‐of‐the‐art computer vision, deep learning, and multi‐sensor technologies. The system combines high‐precision cameras and inertial measurement unit (IMU) sensors to capture real‐time motion trajectories and posture data. These data are processed through convolutional neural networks (CNNs) and long short‐term memory (LSTM) models for accurate movement analysis. Additionally, augmented reality (AR) provides real‐time visual feedback, while deep reinforcement learning algorithms deliver personalized training recommendations. In an 8‐week study involving 100 high school students, the system improved movement accuracy by 23% and received a 4.6/5 user satisfaction rating. Although challenges remain in optimizing computational efficiency and user interface design, the system's modular and scalable nature makes it an adaptable solution for enhancing physical training across educational, home, and professional settings.