AI-Driven Gesture Tracking: A Comprehensive Review of Techniques, Applications, and Future Directions

Abstract: Gesture tracking, the technological ability to interpret human movements as commands, has become a cornerstone of modern Human-Computer Interaction (HCI). Propelled by advances in artificial intelligence (AI), particularly deep learning, gesture recognition systems have evolved from laboratory experiments to practical applications in robotics, manufacturing, healthcare, and consumer electronics. This paper provides a comprehensive review of AI techniques for gesture tracking, spanning the last decade of research. We systematically analyze the gesture recognition pipeline, from data acquisition methods (vision-based, sensor-based) to feature extraction and classification algorithms. The review contrasts traditional machine learning approaches like Support Vector Machines (SVMs) and Hidden Markov Models (HMMs) with modern deep learning architectures, including Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Transformers. A significant focus is placed on training strategies, such as multi-modal fusion and ModDrop, which enhance robustness in real-world conditions. Furthermore, we explore key application domains—Human-Robot Interaction (HRI), Industry 5.0, Augmented Reality (AR), and drone control— highlighting how AI techniques are tailored to meet specific domain challenges. The review concludes by identifying persistent challenges, including occlusion, environmental variability, and the need for large, annotated datasets, and proposes future research directions towards more adaptive, multi-modal, and human-centric gesture recognition systems.

Keywords: Gesture Recognition, Human-Computer Interaction, Deep Learning, Computer Vision, Human-Robot Interaction, Industry 5.0, Multi-modal Fusion