- Use of data mining methods in game analysis in beach volleyball
- Performance analysis in game sports by methods of computer science
- Performance analysis in table tennis
- Injuries in Professional Soccer
- Talent Identification and Development in Soccer
- Performance profile of the national junior squad swimming
- Accuracy of GPS-devices for measuring sport-specific movements
- Goalball
- Automatic image recognition in beach volleyball
- Communication and information flow in complex training systems
- Performance Analysis in Golf Using the ISOPAR Method
- Studying Human Movement Patterns with Self-organizing Maps
- Innovative software for game analysis in beach volleyball
- ASpoGAMo - Automated sport game analysis model
- Qualitative Game Analysis in Handball
- Development and application of a wind tunnel measuring system to improve body positioning in sledging
- Artificial Neural Network Analyses of Human Movement Coordination
Use of data mining methods in game analysis in beach volleyball
Contact: Sebastian Wenninger, Steffen Lang, PD Dr. Daniel Link
Since 2012 game analysis in German beach volleyball has been carried out using two specially developed game monitoring tools. With the BeachScouter, game data can be collected efficiently via a touch surface. The BeachViewer is used for the analysis and presentation of the game data and allows quantitative evaluations to be created and video material to be pre-structured for the qualitative analysis. The previous work was based on the principle of transferring existing expert knowledge into the software. The project aims to use data mining and machine learning methods to extract potentially useful regularities from the existing game data.
Publications:
Wenninger, S., Link, D., & Lames, M. (2020). Performance of machine learning models in application to beach volleyball data. International journal of computer science in sport, 19(1), 24-36. doi: 10.2478/ijcss-2020-0002
Link, D., & Wenninger, S. (2019). Performance streaks in elite beach volleyball - does failure in one sideout affect attacking in the next? Frontiers in Psychology, 10: 919. doi: 10.3389/fpsyg.2019.00919
Wenninger, S., Link, D., & Lames, M. (2019). Data Mining in Elite Beach Volleyball – Detecting Tactical Patterns Using Market Basket Analysis, International Journal of Computer Science in Sport, 18(2), 1-19. doi: https://doi.org/10.2478/ijcss-2019-0010
Promotion video (World Championship Hamburg 2019)
The validity and reliability of the GPEXE-system® was investigated within the scope of an accuracy assessment study. The technology of global positioning systems (GPS) has rapidly advanced over the recent years. Given this convention, team-sport specific GPS-systems have become a common method for assessing the physical demands in professional sports. In order to verify to what extent customary systems are suitable to determine the physical performance under conditions simulating practical use, validation studies are carried out by external providers. GPEXE®, a new entrant in the field of team sport specific GPS-systems commissioned the Department of Training and Computer Science in Sport to undertake an accuracy study. In this respect, the accuracy of the GPS-system was validated by both a laser measurement device (LAVEG®) and a radar system (Local Position Measurement System LPM®). Investigator-in-charge, Daniel Linke, undertook the study within the framework of his dissertation project ‘Performance analysis in team sports based on positional data’.
Research will be conducted into the results of the performance profile of the German Swimming Association within the doctoral project of Christine Hoffmann. These tests include force and flexibility checks on land as well as a multistage endurance test, start and turn analysis and other swimming specific tests in water. On the basis of the newly founded Perspektiv-Team as junior squad of the national team we are interested in establishing a connection between training, test results and competitive performance. First results confirm a reduced base of endurance among the current top athletes compared to the last few decades despite higher swimming speed and higher lactate level.
In the research field “talent” we focus on theoretical concepts of talent identification and development in football. Relative Age Effect, expertise research of top athletes and talent development programmes of elite youth academies and national teams are of particular interest. Goal of the research field „talent“ is the formulation of theoretical concepts regarding talent identification and development as well as the generation of methods in order to assess roster policies in elite youth football. Of special interest in elite youth soccer is the relative age effect (RAE) that emerges from an age-group policy of one year. That policy is responsible for creating chronological age advantages, since there is a one-year difference in age between the youngest and the oldest athletes. This difference in chronological age is referred to as relative age, and its immediate and long-term consequences are known as the relative age effect. One aim of our work is to investigate the impact of the relative age effect (RAE) in German football focusing on the role of the RAE in the transition from youth academy to Bundesliga. A second interesting topic is the approach of expertise research that has already been studied in other domains like chess, mathematics or music. In expertise research it is interesting to study differences of top level players and others, how they reached that expertise and whether there have been any special observations in early ages. Also, the study of possibilities of controlling methods of roster policies in elite youth soccer is of interest. By tracking careers of youth national players and youth Bundesliga players career patterns are generated and help to understand roster policies of youth academies and football associations.
From Epidemiology to Mechanism and beyound
Despite the Bundesliga being one of the top leagues in professional soccer, insight onto the injury characteristic is scarce with only two studies published to date. The aim of the first part of my research program of work is to increase understanding of the epidemiology and injury severity among 1st division Bundesliga soccer players. Specifically, based on information from medial reports, and in addition to traditional variables quantifying injury risk such as incidence-rate, novel dependent variables such as Match-unavilability and Injury-burden will be implemented in a regression model in order to identify playing positions that are of an increase in injury-risk and seasonal variations in injury-risk across the playing season.
The second part of my research project is to empirically test a novel theoretical rational previously published aiming to explore inter-personal interactions between opposing players and the ball leading to contact injuries or high-risk situations in soccer. Specifically, by using a dynamical system theory approach parameters such as inter-personal distance, velocity and angular displacement will be verified, in order to identify key threshold variables by which players might be educated to attune for preventing entering into high injury-risk situations.
In addition, as a physiotherapist, I have serious concerns about the mechanistic approach currently predominate evaluation and treatment following musculoskeletal injuries in clinical practise. This is primarily due to the detachment of the patient from the environment. Addressing this, I am currently working on writing a positional paper highlighting this concern and suggesting how an Ecological Dynamics approach might further enhance the understanding of patient request of help in clinical practise.
Performance analysis in table tennis
Contact: Michael Fuchs, Sebastian Wenninger, PD Dr. Daniel Link,
This project, which is sponsored by the Bundesinstitut für Sportwissenschaft (BISp), is divided in three fields. In the field of theoretical performance diagnostics, mathematical and stochastic models will be checked and extended. The main goal is to develop indices for successful performance in table tennis matches. In the field of practical performance diagnostics methods will be developed and implemented, which support the preparation and follow-up proccesses for competitions efficiently. In the area of methodology of performance diagnostics a customised software for performance analysis will be designed and developed, which takes the requirements and usage conditions of the sport table tennis into account. Because this software influences the work in the other two fields positivly, the main focus at the beginning of the project is on the development of the performance anaylsis software and its implementation for field usage. To have enough material after finishing the software a database will be built up in cooperation with the German Table Tennis Association (DTTB).
The Ph.D. project of Sebastian Wenninger focuses on the application of up-to-date methods of computer science in performance analysis. The goal is to handle the steep rise in available data in this field (i.e. from sensors, computer vision, …) with automated analyses by scientifically founded models and algorithms.
In the first part of the project, stochastic Simulation of tactical behavior in table tennis was realized by numerical derivation. More than 200 world-class table tennis games were modeled as markov chains and automatically evaluated by numerical derivation in order to reveal the influence of specific tactical patterns (i.e. risky/safe play) on the winning probability of a player.
Further studies will utilize methods from the fast-growing fields of Data Mining and Machine Learning to find patterns in large amounts of data. One example is the application of algorithms of market basket analysis to identify (sequential) patterns in the rallies of a beach volleyball game.
Subsequently, so called Long-Short-Term-Memory (LSTM) networks, a specialized version of recurrent neural networks, will be used to establish a connection between particular performance indicators and the resulting success in beach volleyball.
Goalball is the only sport in the Paralympics, other than soccer, for blind athletes. There are three players on each side who try to score by rolling the ball into the opposing net, past the defending team. Each player wears a blindfold so that each player is visually impaired equally. There are bells in the ball so that the players can echolocate it when it is moving. Since Goalball is not a professional sport, there are currently no Goalball specific methods for analyzing performance. This project is foundes by the German Federal Institute For Sport Science (2012-2016) and has developed state of the art software, specific to Goalball (“Goalscout” and “Goalview”). The software operates on tablet PCs for intuitive scouting to be efficient for the user. Furthermore it generates positional data from the video frame which leads to precise information.
Publications:
Link, D., & Weber, Ch. (2018). Finding the Gap: An Empirical Study of the Most Effective Shots in Elite Goalball. PLOS ONE, 13(4): e0196679. doi: 10.1371/journal.pone.0196679
Weber, Ch., & Link, D. (2016). Performance analysis in goalball. In P. Chung, A. Soltoggio, W. C. Dawson, Q. Meng, & M. Pain (Eds.), Advances in intelligent systems and computing 392 (pp. 157–160). Cham: Springer International Publishing. doi:10.1007/978-3-319-24560-7_20
Since 2012 game analysis in German beach volleyball has been carried out using two specially developed game monitoring tools. The aim of this project is to automatically track players and ball positions by means of image detection procedures. If successful, this would substantially reduce effort for coding the video footage, and leave more time for other tasks in the observation, which are currently neglected for time reasons (e.g., block behavior). The project is supported by the German Federal Institute for Sport Science (2012-2014) and is carried out in co-operation with the department of pattern recognition at the University of Erlangen.
Publications:
Gomez, G. Herrera-López, P., Link, D. & Eskofier, B (2014). Tracking of Ball and Players in Beach Volleyball Videos, PLoS ONE 9(11): e111730. doi:10.1371
Link, D. (2014). A Toolset for Beach Volleyball Game Analysis Based on Object Tracking. International Journal of Computer Science in Sport, 13(1), 24-35.
Communication and Information flow in complex training systems
Contact: Raphael Gessenharter, Prof. Dr. Martin Lames
Coaching team sports is a complex and multidisciplinary process, this is particularly evident in professional high performance (PHP) environments. PHP programs reflect the divers and increasingly specialized disciplines that support the modern coaching, training and competition process. Such coaching domains require a distinct perspective for research purposes. The emergence and application of these disciplines, over the last decades has in part been driven by greater scientific interest, increased competition, and economic necessity. Performance improvements go beyond traditional measurable enhancements of individual players such as speed, strength, and accuracy, but also include mental strength, faster (injury) recovery times, injury prevention, and their like. Coaches are expected to develop technical and sport specific skills, communicate and coordinate relevant information effectively within respective youth academies, and plan and prepare for long-term development and participation in high performance competition. While simultaneous keeping track of the all the relevant areas that emerge from the supporting disciplines surrounding PHP teams. Hence the bandwidth of skills required to succeed in such environments spans from physiological, psychological, strategic, technical, and tactical competences to organizational knowledge of the particular club. High performance youth development programs are training programs of professional clubs that comprise all age groups, and game levels. The aim of such youth academies is to identify, form, support, and develop young players to potentially compete on a professional, national, or international level. For professional clubs this task has significantly gained in importance over the last years, where home grown talented players are increasingly part of their long term strategy, and are increasingly a vital source of revenue.
The ISOPAR method provides new opportunities for performance analysis in golf. Based on empirical data the method calculates a topology of average performance with respect to the group of participating golfers. Subsequently, this topology can be illustrated on ISOPAR maps, which are two-dimensional maps of the respective hole with iso-lines superimposed. Iso-lines represent lines of equal number of remaining shots until the ball is holed. Thus, ISOPAR maps are also appropriate to visualize different levels of difficulty on a hole. Furthermore, the ISOPAR method provides the new performance indicators Shot Quality and Shots Saved which account for each unique shot. Shot Quality describes the quality of a shot and Shots Saved describes how a golfer gains or loses advantage on the field with each shot. Using the ISOPAR method, data of the ShotLinkTM database provided by the PGA Tour is analyzed.
Self-organizing maps are a type of artificial neural network useful for visualizing complex human movement coordination. The visualization of the network output can be enhanced by using colour or a third dimension to visualize data clusters, by adding a trajectory to highlight the time-series progression of coordination or by identifying which areas on the output map represent certain critical phases in the movement. There also exists an interesting opportunity to apply this technique to game analyses.
The aim of this project, which is funded by the German Federal Institute of Sports Scoence (2010-2011) is the development and evaluation of two new game observation tools for beach volleyball. The focus lies on optimal information-technological support of the German national teams, particularly in view of the Olympic Games 2012 in London. With the BeachScouter, game data can be gathered in parallel to the contest via touch screen technology. For further analysis the second collection procedure with greater data depth is available. The BeachViewer is used for the analysis and presentation of the game data and allows to create quantitative evaluations as well as to structure video footage for the qualitative analysis. The peculiarity is that characteristic features which had previously to be qualitatively raised by observers can now be automatically calculated from position data.
Publications::
Link, D. & Ahmann, J. (2013). Spielanalyse im Beachvolleyball. Leistungssport, 43(1), 58-63.
Link, D. & Ahmann, J. (2013). Moderne Spielbeobachtung im Beach-Volleyball auf Basis von Positionsdaten. Sportwissenschaft, 43(1), 1-11.
The ASpoGAMo project is funded by the DFG and focuses on the development of a new class of game sport models with respect to soccer. Using positional data of soccer players which are collected using an optical tracking technique, individual movement profiles and performance profiles are created as well as the impact of load on players is studied. Furthermore, the performance of teams is automatically analyzed with respect to actual lineups and teams’ performance in specific game situations. This project is accomplished in cooperation with the faculty of Informatik of the TU München, which programs the respective software.
The project, supported by the BISp deals with the qualitative attendance of the youth national handball team, age-group 1990/91, and the coaches Petersen and Dr. Armbruster by the method of the qualitative game analysis. The scientific innovations are the deployment of intermediation-strategies for the social system of the national team and the empiric proof of effects in game analysis. With this innovations the communication between all stakeholders can be improved. Practically the team got the european championship title 2008 and the junior national team got number two on the world championship in 2007.
Development and application of a wind tunnel measuring system to improve body positioning in sledging
Contact: Dr. Daniel Link
In order to optimize individual body position in sledging a specific measuring system was developed based on the infrastructure of the wind tunnel at the Technical University of Darmstadt. It shows two essential functions: Firstly it contains a prognosis model that allows an estimation of the effects of short changes of the body position on run time. Secondly feedback procedures were integrated that enable one to represent the movement resistance coefficient acoustically through sonification. The measuring system has so far been tested in cooperation with promising athletes at the Olympic training center in Winterberg.
Publication:
Link, D. & Edelmann-Nusser, J. (2012). Drag Coefficient Sonification in Luge Using a Wind Tunnel. International Journal of Sports Engineering and Technology, 6(3), 141-146.
Studies of human movement coordination are complicated by the amount of data collected from modern biomechanical equipment. Self-organizing maps (SOMs) were used in this research to compress redundant information to a useful simpler projection of the data. Movement activities including treadmill running, basketball shooting, golf chipping and running with various orthotic inserts were analyzed using SOMs. From these analyses new characteristics in the data were revealed which provided further validation for the use of SOMs in such studies. New methods were also developed during this project which included various visualization techniques for the output map as well as a technique for training a second map on the output of the first SOM.