Abstract

Abstract

Abstract

Abstract
Nowadays soccer is more than a sport, it is a business industry. Providing financial support to a senior team is a constant challenge for club managers. Flaying this in mind the most relevant factors for a club's life are the game results. They can ensure, amongst other aspects, the continuity of a soccer coach. As a consequence of that, systems supporting decisions represent fundamental tools not only for match' analysis but also to support training sessions. In the literature most academic advances are camera-based solutions and still face some technical issues like occlusion problems and high computational demands. This research project presents new portable and inexpensive soccer analysis software using as base a Wi-Fi real-time tracking system able to identify typical player movements all over a penalty box. The data gathered over a training session constitutes good results especially in what concerns accuracy, thus confirming its robustness and applicability as a tool for scientific soccer analysis and as a valuable support for practical soccer training session. In the future the usage of high-end wireless equipment will increase the system's accuracy and also the coverable area.

Abstract
Over the last few years, research in Artificial Intelligence has focussed on games with incomplete information and non-deterministic moves. The game of Poker is a perfect theme for studying this subject. The best known Poker variant is Texas Hold'em that combines simple rules with a huge amount of possible playing strategies. This paper is focussed on developing algorithms for performing simple online opponent modelling in Texas Hold'em Poker enabling to select the best strategy to play against each given opponent. Several autonomous agents were developed in order to simulate typical Poker player's behaviour and an observer agent was developed, capable of using simple opponent modelling techniques, in order to select the best playing strategy against each opponent. The results obtained in realistic experiments using eight distinct poker playing agents showed the usefulness of the approach. The observer agent is clearly capable of outperforming all their counterparts in all tests performed.

Abstract