Smart Sports Exercises

The Smart Sports Exercises project (2018-2021) was carried out with funding by ZonMw for two years as part of the ZonMw research programme Sports and Movement

This project researched and developed new forms of sports training for volleyball. To this end we worked with a special playing field: an interactive floor that can display graphics and video. In addition, the floor contains unobtrusive pressure sensors.

We developed training systems for this floor that offer tailored, interactive volleyball exercises. The actions and performance of individuals and team can be measured and directly given as feedback. Thanks to the flexibility of these interactive exercises, that can immediately adapt in many ways to player and team, they might be able to train aspects of sports that cannot be addressed without the technology.

In this project we developed a large number of examples of such exercises. Some exercises were only built as non-working example, a movie or similar, to explore the ideas. Other exercises were built completely and could actually be played on the interactive floor. The exercises have been tested with a large number of trainers and athletes. In this way we could improve the technology, but also see whether the technology really has the potential to make a difference. Ultimately, the results of this research should lead to the development of new kinds of “smart sports halls”.

To develop the interactive exercises we also needed to develop technology to automatically recognize actions of players on the floor. The models and algorithms for this have been developed incrementally, for recognizing actions of individual players as well as modelling the pattern of actions between a team during a rally. This was done with volleyball teams of various levels and composition.

The research was carried out by the University of Twente together with Windesheim University of Applied Sciences, Sportservice Veenendaal, InnoSportLab Sport en Beweeg! and the company LedGo – and a huge number of volleyball clubs, teams, trainers, coaches, and athletes.

We started out by looking at volleyball training sessions and interviewing athletes and trainers. We were searching for details of how a training goes, looking for answers to questions such as: what does the team do? Which instructions and feedback does the trainer give? What are often made mistakes?

We then continued in three directions.

Firstly, we looked at the sports perspective. We organised training sessions in which we modified the exercise forms. We use this to further explore the concept of “learning rich training environments”: how can you organise a training session so it has as many as possible opportunities for the player to learn from it? Is “learning richness” something we can influence? Can we measure how “learning rich” a certain training is?

Secondly, we looked at the perspective of interaction technology. We generated ideas for interactive exercises on the smart sports floor and discussed them with athletes and trainers. On the basis of their reactions we are now implementing several game designs that each target different aspects of volleybal training.

Thirdly, we looked at sensor technology. We developed algorithms and models to detect from sensor data which volleyball actions a player is doing at any moment. We look at individual actions (service, smash, set-up, etc) and group actions, and we worked with separate body worn sensors as well as the sensors embedded in the smart floor. The results of this were integrated in order to control the games above.

In dit project worden nieuwe vormen van zaaltraining geïntroduceerd. Dit wordt gedaan in een ‘slimme sporthal’.

Zo heeft het speelveld een interactieve vloer waarop beelden weergegeven kunnen worden. Daarnaast beschikt de vloer over onopvallende druksensoren. Een daarbij ontwikkeld trainingssysteem biedt op maat gemaakte, interactieve oefeningen. Het gedrag en de prestaties van de individuen en het team kunnen direct teruggegeven worden. De oefeningen kunnen daardoor gelijk aangepast worden. De modellen worden verder ontwikkeld door herhaaldelijk versies van het systeem te evalueren. Dit wordt gedaan bij volleybalteams van verschillende niveaus, binnen de sport en het bewegingsonderwijs.