Minecraft Player AI was a project undertaken as part of an extended project qualification at age 17 in College. The title of the piece “History Of Artificial Intelligence In Games And Creating My Own Example” was the main product of the work that was done. At the time my programming skills were limited to basic Java, however looking back this was probably the perfect platform for such a task.
Looking back, I can naturally see many areas I’d have done differently. For example some of the algorithms used for clustering the detected pixels are extremely slow and inefficient however they performed the required task to the levels required.
The project consisted of me trying to emulate the human being trying to play the game as a challenge to see if I could try use the power of computing to be better than the human.
The algorithm was fairly trivial, trying to detect the different colors on the screen clustering them into a decision system that outputs what it thinks it can see on the screen, acting upon such in a feedback manner.
Here is the video I created to log the process :
Here is the final report for the project. It is written in a research paper style, however please don’t laugh too much as I was only 17 when I wrote this! Please also note that the target audience for the report had to be anyone, and thus some sections may seem to talk about extremely trivial problems like basic binary algebra.
I hope to open source the code soon.