A Novel Approach to the Automatic Analysis of Tactics and Actions in Team Sports

Abstract

There can be little doubt that technology has made a contribution towards improving the preparation that professionals who have dedicated their lives to professional sports receive, both from a technical and a performance point of view. This improvement has taken place mainly on an individual level by means of tools that enable, for example, the performance of an elite sportsperson to be monitored. This monitoring process enables highly valuable information to be obtained from data collected by said tools that both the sportsperson and trainer can use for creating a process of continuous improvement. However, from the team sports viewpoint, in which success or failure depends to a large extent on coordinating efforts and using collective tactics and strategies, monitoring actions and behaviour poses a significant challenge. In light of this, an approach based on the automatic detection and analysis of situations and events in team sports has been put forward in this article. In this way, trainers of, for example, a professional football team would have a tool at their disposal which would have the potential to tell automatically if the players behaved according to the tactics and strategies set before a game. To do so, this approach has come to fruition by means of an expert system made up of a reasoning core that uses a knowledge base in which what a player should ideally do according to what has been stated by the trainer, is defined. In this knowledge base, Fuzzy Logic is defined as, a conventionality that allows the way in which human beings think to be represented and drastically bridges the gap there is between the human expert or trainer and machine. The system designed has been used in the specific domain of professional football to detect and analyse situations in which both individual players and the team as a whole are contemplated. The results obtained have allowed the way the players behave on the games field to be automatically assessed according to the knowledge previously passed on to them by the trainer.