Computer Implementation of the board game: Space Hulk

1
Computer Implementation of the board game Space
Hulk
Alexander Michael Spyropoulos
Supervise by Andrew Williams

• An investigation into Artificial Intelligence in
  games

2
Aim

• To determine whether AI complexity affects
  gameplay experience.
• Does an AI have to be more complex in order for
  it to be a challenging opponent?

3
The Test

• In order to answer these questions I have
  produced a simple board game that features 2
  different AI types.
• 1) simple nearest enemy pathfinding
• 2) more complex thinking variety, that checks
  various input data and determines a solution.

4
Artificial Intelligence in Games

• Writing game AI is one of the most challenging
  tasks a programmer can undertake Rabin. S.
  2002 AI Game Programming Wisdom
• Computer controlled opponents have been present
  in many video games since they were first
  developed.
• Paul Tozour states very simple AI was present in
  games such as Pong, Pac-Man, and Space Invaders
  to keep people playing.

5
Space Hulk

• Turn based strategy game, created by Games
  Workshop.
• 2 Player game
• Mission based gameplay

• Space Marines Move Shoot Close Combat
  Overwatch
• Genestealers Move (faster) Close Combat

6
Marine Artificial Intelligence

• Marine AI is governed by 5 main influences

• Distance to the overall mission objective
• Status of other units in the team
• Combat ability
• Currently in combat?
• Distance to nearest enemy

• Some units have unique modifiers

• Flamers are more self conscious, and have
  ammunition concerns
• Sergeant has improved close combat ability over
  other marines

7
Marine Artificial Intelligence

• The Marine AI Phase

• Gathers Data
• Applies data to influence calculation
• Determines Action to be performed (Chooses a
  goal to achieve)
• Feeds the goal into an A path-finding algorithm
• Works out the shortest possible distance to the
  goal.

• Current Stage of Development

• Thought Process completed
• Path-finding Algorithm in production

8
Genestealer Artificial Intelligence

• Genestealers have a much simpler form of AI. They
  are affected by only a few stimuli

• Distance to nearest enemy
• Status of the target unit

• Genestealers rely on speed and sheer weight of
  numbers rather than tactics.

9
Genestealer Artificial Intelligence

• The Genestealer AI Phase

• Looks for nearest target
• Feeds target location into A path-finding
  algorithm
• Works out the shortest possible distance to the
  goal.
• Moves to target and attacks
• Will continue to attack until defeated or target
  is eliminated

• Current Stage of Development

• Currently in the early design stage

10
Game Play

• Here is an early Screenshot of the game in debug
  form.
• At this stage of development only an experimental
  form of the marine AI is running

• By the end of the assignment both AIs will be
  playable, with the player choosing their opponent
  at the start of the game.

11
Play Testing

• The AI will be tested using what Microsoft game
  studios calls the Standard Usability Test
• Players will then be asked to play against both
  AI types, then fill in a questionnaire about
  their experience.
• The results will then be correlated to find out
  which AI is the most popular, challenging, and
  fun to play.

12
Expected Results

• The Marine AI should be more challenging and
  interesting to play, but it is hoped that my
  results could show that you can have just as much
  challenge and fun from the simple AI.

13
Any Questions?
If there are any questions regarding this seminar
or the project itself, please do not hesitate to
ask.
Alex Spyropoulos alexspyropoulos_at_hotmail.com http
//spyrle.sinisterkitten.co.uk