AUTOMATED ANALYSIS OF THE ICE SURFACE CONDITION BASED ON ACCELEROMETER AND GYROSCOPE DATA

Sorokin Vladimir Borisovich

This study proposes a method for automated ice surface quality assessment using accelerometer and gyroscope data from a smartphone. Signals collected during user movement on an ice rink were preprocessed through filtering, segmentation, and feature extraction. Machine learning models—including logistic regression, support vector machine, and random forest—were employed for binary classification of surface condition (smooth/rough). Logistic regression with an optimized classification threshold demonstrated the best performance. The results confirm the feasibility of the proposed method for ice surface monitoring and safety assurance.

Keywords:machine learning, inertial sensors, accelerometer, gyroscope, signal processing, ice surface, classification.