The high performance sports domain, and professional coaching in particular, creates challenges for computer science in sport. Significant multi-disciplinary developments have already been made in the modelling of sports behaviour, computer decision support systems, player tracking technologies, and data mining. Present challenges for the field of Computer Science in Sport (CSS) include issues in mobile computing, multimedia, data visualisation, performance analysis, performance reconstruction, and real time feedback.

A critical role for CSS is to filter and detect meaningful information from vast and diverse sources. Moreover, two significant questions for CSS are: what are the barriers that prevent coaches from embracing sport and computer science, and, how can data be presented in a more meaningful way such that coach expertise is enabled (rather than threatened) by science? The purpose for this Dagstuhl seminar is to connect high performance coaches with the world of computer science, and to bring a modern applied context to computer science work.

Coaches in competitive field games develop expertise using primary modes of game feedback, such as direct visual observation, video review, basic game statistics, input from other first hand observers such as assistant coaches or performance analysts, and crude insights from match outcomes such as the progressive score line. It is from these non-empirical sources that coaches build and test decision making schemes. In practice, coaches become exposed to empirical data, often derived by expert sport scientists, who may provide a potentially bewildering array of performance data from training and games. It may often be that coaches do not possess the knowledge frameworks to absorb or exploit these sources of information. Thus, consideration needs to be given to data visualization, and in particular, matching data presentation techniques to the learning styles, decision making schemes, and game perception frameworks that coaches have otherwise used throughout their careers. This seminar considers how performance data can be processed and analyzed, how analysis results can be communicated and visualized using advances in computer science in ways that enable and amplify coach expertise. The following three sub-themes will be addressed in this seminar.

Computer aided applications (complete hardware and software applications aimed to help coaches)
There is a current trend towards in-situ computing technologies, to acquire, analyze and present performance data without affecting the athletes during training and competition. There are sophisticated feedback systems for coaching to monitor the training process, to optimize performance and prevent athletes from injuries or overtraining. Other feedback systems are using sensors attached to the athlete’s body or to the sports equipment (e.g., accelerometers, gyroscopes, global positioning systems) or sensors embedded in the environment. Important issues are user friendly applications, real-time performance monitoring systems, evidence-based coaching and training, and global interaction and data sharing.

Data acquisition, processing, analysis, and visualization
The integration of modern sensor and communication technologies provides means for developing systems to acquire, process, and transmit data during sports activities. However, the data precision and validity is not always guaranteed. Missing data, wrong and imprecise data needs to be addressed. Moreover, identifying and discarding irrelevant and redundant data and extracting the truly useful data are a major challenge.

Modelling and simulation (interactivity, animations and presentations)
Modelling and simulation aim for developing an understanding of the interaction of the parts of a system, and of the system as a whole. E.g., there are applications such as computational fluid dynamics to optimize swimming techniques, closed loop systems in marathon runners, or even more complex models of passing opportunities and automatic recognition of defensive balance in football. Advances in this topic should provide discussion about key performance indicators to be used in (parsimonious) modelling.

• Arnold Baca (Universität Wien, AT) [dblp]
• Michel S. Brink (University of Groningen, NL)
• Aaron Coutts (University of Technology - Sydney, AU)
• Ricardo Duarte (University of Lisbon, PT) [dblp]
• Björn Eskofier (Universität Erlangen - Nürnberg, DE) [dblp]
• Sofia Fonseca (University Lusófona - Lisboa, PT)
• Wouter Frencken (University of Groningen, NL) [dblp]
• Thomas Jaitner (TU Dortmund, DE) [dblp]
• Peter Lamb (TU München, DE) [dblp]
• Martin Lames (TU München, DE) [dblp]
• Koen A.P.M. Lemmink (University of Groningen, NL) [dblp]
• Daniel Link (TU München, DE) [dblp]
• António Lopes (University Lusófona - Lisboa, PT) [dblp]
• Patrick Lucey (Disney Research - Pittsburgh, US) [dblp]
• Tim McGarry (University of New Brunswick, CA) [dblp]
• Chikara Miyaji (Japan Institute of Sports Science - Tokyo, JP) [dblp]
• Stuart Morgan (Australian Institute of Sport - Bruce, AU) [dblp]
• Jürgen Perl (Mainz, DE) [dblp]
• Jaime Sampaio (Universidade de Trás-os-Montes - Vila Real, PT) [dblp]
• Dietmar Saupe (Universität Konstanz, DE) [dblp]
• Wolfgang Schöllhorn (Universität Mainz, DE) [dblp]
• Thorsten Stein (KIT - Karlsruher Institut für Technologie, DE) [dblp]
• Michael Stöckl (Universität Wien, AT) [dblp]
• Anna Volossovitch (University of Lisbon, PT) [dblp]
• Markus Wagner (University of Adelaide, AU) [dblp]
• Josef Wiemeyer (TU Darmstadt, DE) [dblp]
• Kerstin Witte (Universität Magdeburg, DE) [dblp]