Plan of Action

WorkSheet 1

Matthew McGhie : 0501133
Honours Project Worksheet - #1


State the topic of interest that you may wish to take forward into a project

The topic of interest for this project would be Artificial Intelligence in computer games.

Indicate the issues surrounding this topic that you may wish to investigate

One very important and also rather interesting topic for computer games and in computing in general is artificial intelligence. Over the years this has developed with various new techniques to give the user more realism in game play or interactivity.

In this particular project the focus shall be on a ‘learning’ AI which can be applied to games. By ‘learning’ how a player interacts with the world and NPCs depends on how the AI would react to the player. The AI would observe, learn and counteract with what the player does and plan the actions accordingly.

Various AI techniques should be investigated

Implementation of a learning philosophy for the AI, how the ‘enemy’ store, remembers and interacts with the user depending on player movements.

Possible implementation of a ‘genetic’ memory. The AI continues to learn from interaction and continues to ‘remember’ and grow, either from previous encounters or to another enemy i.e teaching them

The project would also involve using single and group AI, using techniques for a lone ‘enemy’ applying this ‘learning’ ability to plan group ‘attacks’ against the user.
After talking to previous CGT students turned assistants, they told me group AI methods are used and developed by many other students in their projects. While it’s a good technique, you may not want to focus too much on it since markers may be less taken to it after seeing so many others.

If a genetic memory is used could this be indeed used to actually pit enemies against each other with what they have learned from the player, like a self automated learning fight with the victor managing out to out strategise the other.
While an enemy learns and adapts from the player, it may be a concern that each AI member could be continually equal, and waiting for a response rather than making an action right away. Possibly getting a variety of players to pitch in on the original learning process to diversify each enemy technique if the AI was to fight by itself.

Outline your initial ideas about how you would carry out the practical aspects of the project

At the end of the project as with AI a good way to show this is to make a practical demo. This demo would be almost like a ‘game instance’, possibly a basic player and enemy set up. This would allow users to interact with the enemies i.e fight them and from this they would begin to ‘learn’. From this demo you should be able to observe the enemies getting more ‘intelligent’ and more aware of how the user plays. This would also be implemented in a single target and group enemy to the same effect to show the AI in action as it were. Depending on how the project pans out there could even be an option to create and enemy with your own patterns they have learned and pit them against each other in an automated fight where they essentially have to outwit each other.

Initial Ideas

This is basically what i have been thinking about before the new university year began, just what i thought i wanted to do.

Gonna get updated soon, me notes are in another folder just now.