Since the late 1990s, the Robot Soccer World Cup has been used as a testing ground for new technology in the field of robotic design and artificial intelligence. This research initiative pits two teams of robots against each other in a game of soccer. It is hoped that the technology gained will enable the construction of a fully autonomous team of robot players to play a regular soccer game against a human team by the year 2050.
In robot soccer matches, as in real soccer matches, inferring an opponent’s strategy can give a team a major advantage. One important aspect of a team’s strategy is the formation the team uses. Knowing the formations that an opposing team tends to take, enables a team to prepare appropriate countermeasures.
The work presented investigates methods to extract formation information from a completed soccer game. The results show that these methods can be used to infer a classical team formation, as well as other distinguishing characteristics of the players, such as which areas on the field the players tend to occupy, or the players’ movement patterns – both valuable items of information for a future opposition team.
Andre Kriek is a MSc student at Computer Science, Stellenbosch University. His MSc supervisor is Steve Kroon.