Investigating Tilt-Based Technique for Performing Wrist Movement Analysis in Virtual Reality

This paper explores a tilt-based technique for wrist movement analysis using 3D joint position data captured by a standard virtual reality (VR) controller device. This technique leverages natural tilt movements detected by VR controllers, offering a more streamlined and computationally efficient alternative to traditional methods that rely on complex sensor arrays or advanced machine learning models. We conducted a user experiment to evaluate the tilt-based technique in two key areas: movement recognition and kinematic metric measurement. First, we analyzed movement data to precisely recognize six distinct wrist movements, achieving a high F1 score of $93.8 \%$. Next, we assessed the ability of the technique to measure kinematic metrics, specifically focusing on speed and smoothness. Our results showed that particular performance metrics aligned closely with the natural characteristics of the movements.