INTRODUCTION AND MOTIVATION
Cross-country skiing technique has been analyzed using wearable sensors and a dedicated application that transforms movement into objective data. The system has been tested in real environments such as Candanchú, Belagua, Baqueira-Beret and Somport, as well as in controlled conditions at SnoZone (Madrid).
This approach goes beyond traditional visual analysis by combining sensors and data processing to provide measurable insights into performance, helping athletes and coaches improve technique more efficiently.
SPECIFIC OBJECTIVES
Our research focuses on transforming cross-country skiing data into actionable performance insights.
AI Implementation
Development of advanced machine learning models capable of automatically identifying and classifying cross-country skiing techniques with high accuracy.
Cross-Country Analysis Application
A dedicated application that records and synchronizes video with motion data, enabling precise visualization and detailed analysis of skiing technique.
Performance Insights for Cross-Country Skiing
Transformation of data into actionable insights for athletes, coaches and federations, improving performance through objective and data-driven analysis.
MACHINE LEARNING FOR SKI TECHNIQUE CLASSIFICATION
Using IMUs mounted directly on the skis, motion data is captured and processed through a machine learning pipeline to automatically identify different skiing techniques. The system combines sensor data, a custom smartphone application, and deep learning classification to support real-time analysis and provide useful insights for performance evaluation.
Discover more in the full article:
https://www.mdpi.com/1424-8220/24/19/6422
SKI TECHNIQUE ANALYSIS USING SMART INSOLES
Using pressure sensors and accelerometers embedded in smart insoles, the system captures foot pressure distribution and movement dynamics during skiing. These data are processed with machine learning techniques to identify skiing patterns and analyze technique, enabling more precise performance evaluation in real conditions.
Discover more in the full article:
https://www.mdpi.com/1424-8220/25/5/1500
DEVELOPED PROTOTYPE
A complete prototype has been developed to capture and analyze skiing motion in real conditions.
The system integrates inertial sensors (IMUs) mounted directly on the skis, allowing continuous measurement of key variables such as acceleration, pitch and roll angles during movement. These parameters provide a detailed understanding of how the ski interacts with the terrain throughout each stride.
A dedicated application records and synchronizes video with motion data in real time, making it possible to visualize each movement together with its corresponding signals. This provides a much clearer and more intuitive interpretation of the athlete’s technique.
Through this system, it is possible to analyze movement dynamics, propulsion phases and coordination between skis, enabling comparison between left and right sides and supporting the automatic detection of technique patterns.
The goal is to extend this methodology to different disciplines and training environments.
TESTED IN REAL CONDITIONS
Field tests were conducted in real cross-country skiing environments, including routes in Norway. The system was validated down to-22°C, ensuring reliable performance in extreme winter conditions.
Real-Time Cross-Country Skiing Analysis
Our system combines video and sensor data into a single, intuitive application.
IMUs placed on the skis capture motion in real time, allowing each movement to be visualized and analyzed with precision.
Seamless integration between video andmotion data
Real-time visualization and synchronized analysis
AI-powered technique recognition
The system synchronizes data from inertial sensors placed on each ski with video recordings, enabling a precise and intuitive analysis of movement in real conditions.
This approach has made it possible to build a structured database of cross-country skiing techniques, which is used to train artificial intelligence models for automatic classification.
The figure illustrates an example of synchronized signals and motion data, showing how different movement phases and technique patterns can be identified.
