Automated identification and evaluation of subtechniques in classical-style roller skiing

The aims of the present study were (1) the development of an automated system for identifying classical-style ski subtechniques using angular rate sensors, and (2) the determination of the relationships among skiing velocity, ski course conditions, and ski subtechniques using a global navigation satellite system (GNSS) and the developed automated identification system. In the first experiment, the performance of a male cross-country skier was used to develop an automated system for identifying classical-style ski subtechniques. In the second one, the performances of five male and five female college cross-country skiers were used to validate the developed identification system. Each subject wore inertial sensors on both wrists and both roller skis, a small video camera on the helmet, and a GNSS receiver. All subjects skied a 6,900-m roller ski course using the classical-style at their maximum speed. The adopted subtechniques were identified by the automated method based on the data obtained from the sensors, and also by visual count from a video recording of the same ski run. The results showed that the automated identification method could be definitively used to recognize various subtechniques. Specifically, the system correctly identified 9,307 subtechnique cycles out of a total of 9,444 counted visually, which indicated an accuracy of 98.5%. We also measured the skiing velocity and the course slope using the GNSS module. The data was then used to determine the subtechnique distributions as a function of the inclination and skiing velocity. It was observed that male and female skiers selected double poling below 6.7° and 5.5° uphill, respectively. In addition, male and female skiers selected diagonal stride above 0.7° and 2.5° uphill, and below 5.4 m/s and 4.5 m/s velocity, respectively. These results implied that the subtechnique distribution plot could be used to analyze the technical characteristics of each skier. Key PointsThe automatic identification method, which utilizes data obtained by small and light inertial sensors, could be used to recognize subtechniques of classical-style roller skiing with a high accuracy of 98.5%.The skiing velocity was measured using a small DGNSS module at all over the course, which made it possible to evaluate the technical features of skiers together with the results of the automatic identification.However, there were limitations in the automatic identification during the start phase, the downhill, and the transition period between subtechniques.