Este trabalho descreve a implementação de um sistema multi agente usando Redes de Comportamento configurada por Algoritmos Genéticos. O sistema utiliza o jogo de computador Quake II como o ambiente simulado para os agentes. Redes de Comportamento são utilizadas como o mecanismo de tomada de decisão. Um Algoritmo Genético é utilizado para configurar os parâmetros da Rede de Comportamento. Cada agente é um programa independente que se conecta ao servidor do jogo para realizar tarefas e trocar material genético a fim de evoluir. Os resultados obtidos mostram um ambiente multi agente dinamicamente configurado capaz de evoluir e se adaptar apropriadamente conforme o andamento do jogo. / This work describes the implementation of a multi-agent system using Behavior Networks configured by Genetic Algorithms. The system uses the computer game Quake II as the simulated environment for the agents. Behavior Networks are used as the decision making mechanism. The Genetic Algorithm is used to configure the parameters of the Behavior Network. Each agent of the system is an independent program that connects to the game server to perform tasks and to exchange genetic material in order to evolve. The results obtained indicate a dynamically configured multi-agent system that can evolve and adapt accordingly throughout the course of the game.

Este trabalho descreve a implementação de um sistema multi agente usando Redes de Comportamento configurada por Algoritmos Genéticos. O sistema utiliza o jogo de computador Quake II como o ambiente simulado para os agentes. Redes de Comportamento são utilizadas como o mecanismo de tomada de decisão. Um Algoritmo Genético é utilizado para configurar os parâmetros da Rede de Comportamento. Cada agente é um programa independente que se conecta ao servidor do jogo para realizar tarefas e trocar material genético a fim de evoluir. Os resultados obtidos mostram um ambiente multi agente dinamicamente configurado capaz de evoluir e se adaptar apropriadamente conforme o andamento do jogo. / This work describes the implementation of a multi-agent system using Behavior Networks configured by Genetic Algorithms. The system uses the computer game Quake II as the simulated environment for the agents. Behavior Networks are used as the decision making mechanism. The Genetic Algorithm is used to configure the parameters of the Behavior Network. Each agent of the system is an independent program that connects to the game server to perform tasks and to exchange genetic material in order to evolve. The results obtained indicate a dynamically configured multi-agent system that can evolve and adapt accordingly throughout the course of the game.

Vortrag zum 135. Unix-Stammtisch * Das Sourceforge-Projekt Jake2 * Wie langsam ist Java wirklich? * 3D-Game-Engine Interna

No description available.

PC panel strengthening technique was developed in M.E.T.U. Structural Mechanics Laboratory in order to respond the need of practical and efficient pre-quake seismic strengthening procedures applicable to RC framed structures. The idea behind the method is simply to convert the non-structural infills into load bearing structural elements by gluing PC panels over the existing infill wall surface. The remarkable advantages of the procedure is not only the considerable amount of seismic performance improvement but also the simplicity of application, very low levels of disturbance to the occupants and most importantly, the applicability during service. A number of PC panel application parameters were experimentally investigated by previous researchers. The success of PC panel method on seismic performance improvement of RC frames with different aspect ratios was experimentally investigated in the present study. Total of fifteen, 1:3 scaled, one-bay, two-storey RC frames were tested in three various aspect ratio series. Constant axial load was applied to the columns and reversed cyclic load was applied in the lateral direction. Hollow brick v infilled frame and cast-in-place RC infilled frame were the lower and upper bound reference specimens, respectively. Seismic performance indicators such as response envelope curves, lateral load carrying capacities, cumulative energy dissipations, initial stiffness indicators and ductility values clearly showed the effectiveness of PC panel application over different geometry of RC frames of concern. Moreover, PC panel application either with rectangular or with strip shaped PC panels provided seismic performance improvement to be almost equal to that of cast-in-place RC infill application for all series. Equivalent diagonal strut concept was followed in analytical studies to simulate the infills of RC frame openings. The required strut material properties were estimated from total of eighteen individual wall panel tests. The bond-slip effect, due to utilization low strength of concrete and plain rebars, was also investigated and introduced to the analytical frame models. Non-linear push over analysis was performed for all specimens in OpenSees computer software. The analytical results were compared with that of experimental response envelopes.

Vortrag zum 135. Unix-Stammtisch * Das Sourceforge-Projekt Jake2 * Wie langsam ist Java wirklich? * 3D-Game-Engine Interna