Design and Development of an Automated Gymnastics Movement Evaluation Device

Abstract

Rhythmic gymnastics need precision, balance, and harmonious body coordination. This study designs an automated rhythmic-gymnastics movement evaluation device based on wearable inertial and orientation sensing. An ADXL335 accelerometer sensor and an HMC5883L magnetometer sensor are integrated with an ESP32-C3 mini transmitter to acquire three-axis acceleration and heading data in real time and stream them via Wi-Fi (UDP) to a Wemos D1 Mini ESP32 as a receiver. The receiver sends data to a PC application that performs downsampling with average pooling to reduce signal density, applies min–max normalisation to harmonise scale across axes, and compares the processed test sequence against reference data. Deviations are quantified through per-parameter error computation, which is converted to accuracy and combined using predefined weights to generate an overall performance score that is displayed on an LCD/P10 panel. Experimental trials on SKJ 2012 warm-up movements show that sequences closely matching the reference pattern consistently yield high scores about 80–85, whereas intentionally incorrect executions produce substantially lower scores around 40. The proposed system therefore provides an objective, repeatable, and real-time scoring mechanism that can support teachers in evaluating movement conformity, reduce assessor bias, and enable measurable feedback for learning and training in school-based physical education.