Desenvolvimento de protótipo para identificação de golpes de membros inferiores do Muay Thai

Abstract

This work presents a study on the development of a prototype for identifying Muay Thai kicks as a way to aid in training for athlete progression. The motion sensor of Meta Motion R of the company MBIENTLAB was chosen, which already contains Accelerometer, Gyroscope and Magnetometer. The sensor fits into the MEMs category which are Microelectromechanical Systems technologies, it integrates electrical, mechanical, sensor and electronic elements into a single device. This device has the characteristic of being very small and has the possibility of being programmed for the desired purpose, which allows the advantage of a quantitative sample of the performance in the accomplishment of movements within the scope of the sport activity, which brings the collected data to a more realistic exercise of the sport. After conducting a study on the Muay Thai modality, three different types of kicks were chosen for the experiment: low kick, back kick and frontal kick; for being common kicks with differences between them, as the execution plan and acceleration. Having established the type of movement, different samples were collected, with the sensor coupled to the athlete’s kicking leg. Using a Kalman filter already embedded in the sensor, the measured data was sent wirelessly to a system that, through Artificial Neural Networks, was treat the measures received and identify the movements ofthe fighter. The results for low kick and back kick were satisfactory, however, for the frontal kick there was a considerable decrease in the success of the identification. This discrepancy of results can be related to the technique of movement, where the low kick and back kick are rotating kicks and the frontal is a straight kick which can cause a great difference in the data mainly of the gyroscope that affects angular velocities and thus impaired the RNA and the experiment.