Abstract
Soccer is a team sport in which the performances of all team members are important for the outcome of a match. Even though the analysis of game events can be used to measure the team's performance, their perception, especially during the match, is extremely difficult, even for the involved agents. Soccer has been used as a simulation environment in many studies, mainly in the area of robotics. The RoboCup is an international robotics competition with an ambitious goal: in 2050 a robotics team will be capable of defeating the human world champion at the time. In this context, we compared technical similarities between human and robotics soccer. Based on an off-line automatic event detection tool, game statistics for the finals of both human and robotics soccer tournaments were collected and compared using the Wilcoxon test. The results show that the most frequent event in both forms of soccer is successful passes. Analysing the two types of passes considered (successful and missed), we conclude that there are significant differences between the two forms (W = 2, P = 0.000354), with human soccer presenting a higher percentage of successful passes (77.89% vs. 66.97%). Of restart events (W = 0, P = 0.00048965), the most frequent one, in both forms, is the throw-in (human 59.91%, robotics 66.4%), and the least frequent is the corner (human 13.7%, robotics 14.09%). Regarding the frequency of shots, in the robotics environment "shots" were the most predominant type (43.27%), whereas in human soccer "shots on target" predominated (71.25%; W = 64, P = 0.000085641). Finally, the number of faults is minor in robotics soccer.

Abstract
To get the most out of powerful tools, expert knowledge is often required. Experts are the ones with the suitable knowledge to tune the tools' parameters. In this paper we assess several techniques which can automatically fine-tune ANN parameters. Those techniques include the use of GA and stratified sampling. The fine-tuning includes the choice of the best ANN structure and the best network biases and their weights. Empirical results achieved in experiments performed using nine heterogeneous data sets show that the use of the proposed Stratified Sampling technique is advantageous. © 2009 Springer Science+Business Media, LLC.

Abstract
This paper presents a hybrid behaviour process for performing collaborative tasks and coordination capabilities in a rescue team. RoboCup Rescue simulator and its associated international competition are used as the testbed for our proposal. Unlike other published work in this field one of our main concerns is having good results on RoboCup Rescue championships by emerging behaviour in agents using a biological paradigm. The benefit comes from the hierarchic and parallel organisation of the mammalian brain. In our behaviour process, Artificial Neural Networks are used in order to make agents capable of learning information from the environment. This allows agents to improve several algorithms like their Path Finding Algorithm to find the shortest path between two points. Also, we aim to filter the most important messages that arise from the environment, to make the right choice on the best path planning among many alternatives, in a short time. A policy action was implemented using Kohonen's network, Dijkstra's and D* algorithm. This policy has achieved good results in our tests, getting our team classified for RoboCup Rescue Simulation League 2005.

Abstract
This paper presents the AFRANCI tool for the development of Multi-Strategy learning systems. Designing a Multi-Strategy system using AFRANCI is a two step process. The use interactively designs the structure of the system and then chooses the learning strategies for each module. After providing the datasets all modules as automatically trained. The system is aware and takes into consideration the inter-dependency of the modules. The tool has built-in learning algorithms but can use external programs implementing the learning algorithms. The tool has the following facilities. It allows any user to design in an interactive and easy fashion the structure of the target system. The structure of the target system is a collection of interconnected modules. The user may then choose the different learning algorithms to construct each module. The tool has several built-in Machine Learning algorithms has has interfaces that enables it to use external learning tools like WEKA and CN2. AFRANCI uses the interdependency of the modules to determine the sequence of training. For each module the system uses a wrapper to tune automatically the parameters of the learning algorithm. In the final step of the design sequence AFRANCI generates a compact and legible ready-to-use ANSI C open-source code for the final system.

Abstract
This paper presents the PyrantidNet Tool its a fast and easy way to develop Modular and Hierarchic Neural Network-based Systems. This tool facilitates the fast emergence of autonomous behaviours in agents because it uses a hierarchic and modular control methodology of heterogeneous learning modules: the pyramid. Using the graphical resources of PyramidNet the user is able to specify a behaviour system even having little understanding of artificial neural networks. Experimental tests have shown that a very significant speedup is attained in the development of modular and hierarchic neural network-based systems when using this tool.

Abstract