Detection of Different Throw Types and Ball Velocity With IMUs and Machine Learning in Team Handball

van den Tillaar, R
Bhandurge, S
Stewart, T
Item type
Conference Contribution
Degree name
Journal Title
Journal ISSN
Volume Title
International Society of Biomechanics in Sport (ISBS)

The purpose of this study was to investigate if an inertial measurement unit (IMU) and machine learning could be used to detect different types of team handball throws and predict ball velocity. Throwing was measured using IMUs and a radar gun in seventeen participants during standing, running and jump throws with a circular and whip-like wind up. Using these data, machine learning could predict peak ball velocity with an error of 1.05 m/s and classify approach types and throw types with ~85–90% accuracy. It was concluded that to monitor throwing load, the combination of inertial measurement units and machine learning offers a practical and automated method of quantifying throw counts and discriminating throw types in handball players under standard conditions.

Throwing velocity; Artificial intelligence
ISBS Proceedings Archive: Vol. 38 : Iss. 1 , Article 48. Available at:
Rights statement
The following uses are always permitted to the author(s) and do not require further permission from provided the author does not alter the format or content of the articles, including the copyright notification: Posting of the article on the internet as part of a non-commercial open access institutional repository or other non-commercial open access publication site affiliated with the author(s)'s place of employment.