Training Science: Artificial Intelligence Can Predict Handball Goals

Status: 18.07.2022 16:43

In a challenge with SG Flensburg-Handewitt, researchers from the University of Paderborn developed synthetic intelligence that may predict handball targets. They additionally developed algorithms to scale back the danger of damage and enhance the athlete’s efficiency.

Mini-computers worn on the physique – wearable units – have lengthy been a part of the principle gear in aggressive sports activities. The units generate infinite quantities of information and supply details about athletes’ health and stress ranges. But solely through the use of synthetic intelligence (AI) can the quantity of information obtained be used optimally. However, thus far, AI-powered information evaluation in aggressive sports activities has been restricted, even with worldwide skilled soccer heavyweights equivalent to Liverpool FC or Barcelona FC pioneering.

Injury danger and efficiency components

“The information was systematically recorded within the German handball league for 2 years,” explains prof. Jochen Baumeister, Head of the Working Group on Learning and Neuroscience on the University of Paderborn. However, widespread use of the corresponding information remains to be distant. Baumeister’s group and his colleague from Paderborn, enterprise informatics specialist Prof. Dr. Oliver Müller, along with the skilled handball group SG Flensburg-Handewitt, a number one European membership, investigated how the usage of algorithms can cut back the danger of damage and enhance the efficiency of athletes.

AI “quantifies” efficiency

In addition, Paderborn researchers have even developed a man-made intelligence that may predict targets! As described by Flensburg athletics coach Michael Doering, “particular person performances [der] Players have been “digital”. But how does it work? Called AI, PIVOT makes use of sensors that document the place of gamers and the ball each 50 milliseconds to independently acknowledge motion patterns. Using the acquired information, the AI ​​can predict the likelihood of scoring a aim within the subsequent few seconds.

Target likelihood calculated in actual time

“PIVOT lets you calculate the likelihood of scoring or conceding a aim in actual time. If you examine this likelihood earlier than and after a sure sport, you possibly can measure how good or dangerous this sport was and the gamers’ choices primarily based on it,” says information evaluation professional Muller. explaining the precept. His Paderborn colleague Baumeister added: “Quantitative measures of efficiency are an excellent foundation for finding out connections such because the nervous system. It offers invaluable insights into the state and actions of gamers.”

For now, solely the ball-carrying participant is in focus

So far, Paderborn’s analysis group has solely evaluated the conduct of gamers carrying the ball. According to Baumeister, the extension can present extra readability to gamers who haven’t got the ball. For every state of affairs, the distinction between the expected very best conduct and the precise conduct may also be calculated.

In any case, SG Flensburg-Handewitt coach Mike Machulla sees fully new prospects and higher help in each day coaching and video games from the brand new findings of “modern information processing”.

Links/Research

(d)

.

Leave a Comment