Recognition and Control of the Athlete's Movements Using a Wearable Electronics System

The paper presents a wearable electronics system and an algorithm for recognizing and controlling the athlete's movements accuracy. The proposed system is composed of a sensor network, which comprises a microcontroller, accelerometers, myosensors and a heart rate module. The presented motion recognition algorithm consists of signal acquisition, signal preprocessing, feature extraction and normalization, dynamic human motion (exercise) detection and recognition, and the exercise accuracy control. The proposed algorithm is based on the use of modern machine learning methods, in particular, on the basis of recurrent neural networks. As an experiment, basic exercises were used: push-ups, squats and the deadlift. Experimental results showed that our approach could reach a 97% recognition level as well as make it possible to detect errors during the exercising.