Sistema multiagente para la simulación de la defensa de un edificio ante un ataque enemigo
Cargando...
Fecha
Autores
Rodríguez Abed, Abdel
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central “Marta Abreu” de Las Villas
Resumen
En este trabajo se hace el diseño y la implementación de un sistema multiagente para la simulación de la defensa de un edificio ante un ataque enemigo. Para esto se hace un estudio de las estrategias de simulación, los sistemas multiagentes y del aprendizaje reforzado como técnica de cooperación entre agentes. Se hace una revisión bibliográfica de la aplicación de estas técnicas en problemas reales. La simulación se basa en una estrategia sincrónica por intervalos de tiempo. Los agentes se diseñan como sistemas basados en reglas. Su cooperación se realiza mediante especialización y comunicación. Esta última se logra a través de aprendizaje reforzado y envío de mensajes. Para la implementación del sistema multiagente se diseña e implementa una jerarquía de clases con características extensibles para otros problemas que se ajusten a este modelo. Se implementan además objetos para la interfaz visual usando OpenGL.
In this work we design and implement a multi-agent system to simulate the defense of a building given an enemy attack. A study of the simulation strategies, the multi-agent systems and of the reinforcement learning as cooperation technique among agents is made. Besides, a bibliographical revision on application of these techniques in real problems is made. The simulation is based on a synchronous methodology by time intervals. The agents are designed as rules based systems. Their cooperation is carried out by means of specialization and communication. This last one is achieved through reinforcement learning and messages sending. To implement the multi-agent system it is designed and implemented a class hierarchy with characteristic expandable for other problems with this model. Some objects to the visual interface using OpenGL are also implemented.
In this work we design and implement a multi-agent system to simulate the defense of a building given an enemy attack. A study of the simulation strategies, the multi-agent systems and of the reinforcement learning as cooperation technique among agents is made. Besides, a bibliographical revision on application of these techniques in real problems is made. The simulation is based on a synchronous methodology by time intervals. The agents are designed as rules based systems. Their cooperation is carried out by means of specialization and communication. This last one is achieved through reinforcement learning and messages sending. To implement the multi-agent system it is designed and implemented a class hierarchy with characteristic expandable for other problems with this model. Some objects to the visual interface using OpenGL are also implemented.