NUI Galway – UL Alliance First Annual ENGINEERING AND - ARAN ...

Generation and Analysis of Graph Structures with an Application to Generate
Levels (Turn Based Patroller-Intruder Games)
David Power and Colm O’Riordan
CIRG, NUI, Galway
d.power1@nuigalway.ie, colm.oriordan@nuigalway.ie
Abstract
This project focuses on automatically testing game
content and consequently designing a tool which
automatically generates game content. This research is
relevant to current game design techniques as it
reduces game testing times and allows for quick
generation of content.
1. Introduction
Current game design fundamentals include
conceptualising a game, planning how this shall be
accomplished, executing this plan, testing the results
and then refining the game [1]. This can be a long
process along with extensive testing needed to be
performed before a game can be completed. One way
for the testing time to be reduced would be to develop
automated testing of the game that could occur during
all levels of the design stage.
Another way to decrease the length of game design
would be to automate level design. This is especially
useful for genres like patrol games, FPS (First Person
Shooter), RTS (Real Time Strategy), where levels are
often quite similar with just small changes in terrain
allowing for different gaming experiences. If the
properties of a level were analysed, then levels could be
generated automatically allowing for much more game
content, increasing longevity and enjoyability for users.
In this research, the domain of patrol games is used
develop ways of testing levels automatically and
generating levels automatically based on the results. A
patrol game involves an intruder agent and one or more
patroller agents [2]. The intruder is inserted into a
predefined area in a level and then must make its way
towards another predefined area in the level known as
the goal. The patrollers are constantly patrolling the
level on predetermined paths trying to find the intruder.
An intruder success is when it reaches the goal area
while a failure occurs when a patroller and an intruder
cross paths.
2. Simulator Model
The graphs being tested by the simulator correspond
to levels of a 2D turn based, node based, patrollerintruder
game. A level is represented as an adjacency
matrix and all possible pathways through the graph are
calculated by brute force. The simulator then randomly
generates paths for both the intruder and patroller. The
simulator plays the two agents against each other and
records either a success or failure depending on the
outcome. If the result is a failure, the positions of the
3
agents at failure are also recorded. The agents are
competed against each other for a large sample of
randomly generated paths. The ratio of successes to
failures will show the level of difficulty of the graph.
3. Current Work
By recording the frequency of failure localised to
each node with respect to the total number of failures
occurring throughout the simulation we hope to identify
certain graph characteristics. The levels of difficulty
each of these characteristics bring to a graph need to
documented and assigned and influence value.
The main characteristics identified so far that
influence graphs are connectivity, choke points, chains,
dead ends and small cycles. Choke points for instance
severely restrict pathways through an area of the maze
making safe navigation for the intruder more difficult.
4. Future Work
Using the values described above, we hope to be
able to generate new graphs within a certain level of
difficulty. This will be accomplished by evaluating
different sections of graph and using genetic algorithms
to form a totally new graph from constituent parts.
Furthermore we hope to run these newly generated
graphs through another simulator (3D real time game)
created by a colleague [3]. By comparing the results of
the two simulators we will see if the same
characteristics and values identified in the 2D turn
based environment, hold true for the 3D real time
environment.
5. References
[1] Chris Crawford, The Art of Computer Game Design,
Osborne/McGraw-Hill, Berkeley, CA, 1984
[2] Amigoni, F.; Basilico, N.; Gatti, N.; Saporiti, A.; Troiani,
S.; "Moving game theoretical patrolling strategies from theory
to practice: An USARSim simulation," Robotics and
Automation (ICRA), 2010 IEEE International Conference on ,
vol., no., pp.426-431, 3-7 May 2010
[3] Costello, F. and O’Riordan, C. “An Approach to Providing
Feedback at the Design Phase in Game Authoring Tools”.
GAME-ON 2009: 10 th International Conference on Intelligent
Games and Simulations (ed. Linda Breitlauch). Mediadesign
Hochschule Dusseldorf and EUROSIS, pp 20-23.