Artificial intelligence as a tactics assistant in football

Could artificial intelligence revolutionize football? At least according to Google. Because the Untenrhemn developed the AI ​​system TacticAI together with Liverpool FC. In the future, the tactics assistant will act as an advisor for corner kicks.

Google wants to get more out of corner kicks in football in the future. The AI ​​assistant TacticAI, which researchers developed together with Liverpool FC, is intended to help.

The US company has the AI ​​assistant in the scientific open access journal Nature Communications presented. This is intended to provide tactical insights through predictive and generative AI – especially for corner shots in football.

TacticAI is being developed at Liverpool FC

Google’s AI researchers worked with Liverpool FC to develop TacticAI. In the future, the system will enable coaches to test alternative player lineups for each routine.

The trainers were probably quite happy with that. According to Google, they preferred the AI ​​assistant's suggestions over practical tactical formations in 90 percent of cases.

Sports like football are also a dynamic area for AI development as they involve real-world interactions between multiple agents with multimodal data.

The researchers behind the AI ​​assistant want to answer three core questions in particular with TacticAI. This includes, among other things, predicting who the ball is most likely to land in a corner kick in a particular formation.

But TacticAI should also be used after the game. The second research question revolves around game analysis and the question of whether similar tactics have worked well in the past.

The third question now relates to the learnings from these evaluations. The researchers want to use TacticAI to determine how tactics need to be adjusted to achieve a specific result.

Geometric deep learning predicts corner kicks

TacticAI uses a geometric deep learning approach to predict corner kicks in football. To do this, the tactics assistant displays the corner kick formations as a graph.

The nodes represent the players, who are also assigned characteristics such as position, speed or height, among other things. The edges, in turn, represent the relationships between the players.

Predicting the outcomes of corner kicks is very complex due to the randomness of play by individual players and the dynamics between them.

This allows TacticAI to find similar corner kicks and help coaches test different tactics. But TacticAI can also support the redesign of tactics for a corner kick.

The AI ​​system provides tactical recommendations for the position of all players on a team. This allows coaches to recognize important patterns and key players more quickly.

Also interesting: