Wednesday, March 6, 2013

Dissertation - Formulating Ideas

At the end of this semester I will be submitting a proposal for my MSc dissertation. The dissertation will be a semester long project during which I will have no other coursework, and will be working full time researching a game programming topic. I will be expected to present new and useful findings that better the game industry knowledge as a whole, with the understanding that this is not a PhD and I only have one semester in which to complete the research.

Today, I've at last settled on a possible research topic. As regular readers will remember, last semester I created a fuzzy logic controller and genetic algorithm optimization using HTML5 JavaScript. My lecturer Dr. David King seemed interested in developing the project further, but I wasn't sure how to go about doing that. After our meeting today, however, I think I have found a way that combines many of my programming interests into one project.

Fuzzy controller input membership functions for last semester's project

I'll frame the topic first as a product: An online tool that game developers can use to create fuzzy control systems for their games and automatically generate a plugin that is ready to use. Optionally, it can also generate a genetic algorithm optimization method for them - however, since this requires knowledge of the game implementation, this would be more difficult to implement - but also more interesting and potentially rewarding.

However, this is not just a semester long project - it is a research dissertation. I have to be able to phrase my topic in terms of an investigation, or research question. I'm still figuring out how to do that, but one possibility is: "Can a modular FIS tool be created to suit an arbitrary game's development needs, and if so, how useful is it to the development process?" Measurables include: How effective the FIS generated by the project are (do they work); how easy the interface is to use (survey); and how easy the generated plugins/modules are to use in the final game (survey). Things that would be needed to show this include: the tool itself (HTML/JavaScript); and one or more game applications that use the generated plugins, if possible some games that weren't created by me. I may be able to get Dr. King to make the tool available to next year's AI module for MSc students to get some feedback and help them understand FIS controllers, and may also release the tool to the Abertay Game Development Society in return for feedback and use of their game in my dissertation demonstration.

My next task is to research and see if this has been done in any form previously. I'll also need to do general background research into fuzzy controllers  genetic algorithms, writing plugins and tool programming, and user interface design. Once the base research has been done, I'll write my proposal and get started.

While I'm a little concerned that this project may not qualify as research, and more just making something, I'm hoping it can be massaged into meeting the assignment requirements. It is a great combination of my interests, in AI, tool programming, and user interface/interaction. I may be creating something that is immediately useful to fellow students and developers. I'm quite excited and looking forward to developing the idea further.

No comments:

Post a Comment