[2E1-GS-10] AI application:

[2E1-GS-10-01] Penalty kick prediction using time series data and Echo State Network

〇Shuya Amano¹, Ryosuke Hosaka¹ (1. Shibaura Institute of Technology)

Keywords:time series data, football, Echo State Network

The goal of this research is to have AI learn the kicking form of penalty kicks, predict whether the ball will fly to the left or right, and obtain the basis for this decision. In penalty kicks, the time between when the ball is kicked and when it reaches the goal is very short, so goalkeeper needs to make predictions based on the kicking form. However, We had the impression that predictions were mainly empirical and lacked scientific basis. Therefore, we thought that using AI to make predictions would provide a useful basis for goalkeepers to make predictions. Past research had attempted to use CNN-based methods. However, this method does not sufficiently consider the relationship between time series data. Therefore, in our research, We used an Echo State Network (ESN), a type of reservoir computing, to make the time-series data related to each other and to focus on learning movements. As a result, predictions were made with an accuracy of more than 95%. In the future, we plan to work on extracting evidence in the predictions and adapting the predictions to more complex situations.