Recognizing Attack Patterns: Clustering of Optical Flow Vectors in RoboCup Soccer

In contrast to human soccer players, autonomous robot soccer players often move according to a limited set of predefined behavioural rules. This knowledge can be used advantageously: If the opponent’s behavioural rules are learned, it will be possible to detect these during a match and react accordingly. A method for autonomous activity mining in videos, called Probabilistic Latent Sequential Motifs, is used to find optical flow patterns in video’s of a robot soccer player during a penalty shootout. The found patterns are used by a humanoid goalkeeper to predict and anticipate opponent behaviour. Effectiveness of the method is tested by comparing the performance of this goalkeeper, i.e., the ratio of number of goals to number of goals prevented, to that of an existing goalkeeper that only reacts when the ball approaches at sufficient speed. Results show that the found goalkeeper performs fairly well, but that it loses to the existing goalkeeper in terms of performance. Methods that may improve performance are discussed.