Developement of Intelligent Agents for the AgentGame Environment

OData support
Eredics Péter
Department of Measurement and Information Systems

This thesis gives a brief introduction about the AgentGame, which based on the Jason framework. In the game two teams competing against each other, whose goal is to maximize the consumes foods with their agents.

The thesis is designed to stretch the AgentGame boundaries and to find an optimal solution to the emerging problems . To reach the limits of the system I need to make some changes which changed the game play fundamentally. I built into the game the possibility to change the map size periodically. To leave the game enjoyable even after these changes I need to change the number of meals.

This paper elude the optimal paths traversing constant size maps. To find the optimal algorithm, necessary to introduce the concept of utility and need to choose the correct algorithm for generating a series of steps .

For the dynamic-sized map traverse I have created a complex algorithm and it’s pay attention for the position of agent , the enemies food situation and the track edge.

To increase the speed of this method I used the macro block, I observe the edges of the map and infulence the number of the turns and also I take care for the harmony between the running time and the performance.In addition, I put the emphasis on how the agents approach the food.

For testing these principles, techniques what described, I created an interface and automatizated the testing, after I analyzed the outcomes.

At the end of the document I describe the functionality and the potential opportunities, useage and limitations of the algorithm.


Please sign in to download the files of this thesis.