Pokemon AI

[Andy-Olshanky]

A few years back, my friends and I made a rudimentary Pokemon battling program along with a very rudimentary battling AI that we never properly finished. I propose making a properly working battle AI using TensorFlow as well as implementing more Pokemon into the database as well as a proper UI that shows move selection, pokemon, and some simple battle animations. Currently, the battling is done by printing to the console and using console inputs, but a proper UI would make the experience much better overall. This would also be written in python, and here is the link to the project github: https://github.com/Andy-Olshanky/pokemon-ai

[PatrickBrady7]

I chose this proposal as one of the best because I think with the right team it would be a fun passion project. I'm sure there's enough people in this class to assemble of team of people who love pokemon or are very familiar with the games. Also having people who understand how the ai has worked in the games would obviously help with how to code the ai's behaviors. I think a great addition to this would be to have difficulty modes or the ability to play against ai that are functionally different (something that's already in the base game when it comes to "harder" fights).

[ruuffian]

I've enjoyed a few recent Pokemon-related games, but I've never actually played all the way through the older games. Regardless, I think this is an interesting project. Using a ML model for a battle AI sounds like a fun challenge, and porting the cmdline app to a fully functioning standalone GUI will definitely be a lot of work, but develop some skills I haven't used before.

Right now, to create a new model for an opponent AI, the user needs to know Python. If the TensorFlow AI model goes well enough, it could be interesting to explore some kind of GUI for users to create their own 'AI', perhaps even allowing them to battle each other as a kind of autochess (if you're familiar with TFT, think that but with pokemon rather than League). This may be worth it's own project entirely, but could be interesting to explore regardless.

[gregory-shakarjian]

I prefer this project because I have an interest in both AI and Pokemon. I've always wanted to work on implementing my own coding into a game, so this would be a great start! Additionally, getting more experience with AI is something I'm looking for, so this is also a good start. It also seems like there has been a lot of work put into this project already, so there would be a lot to work with. Overall, this fits my interests and seems like a good way to get experience.

While I don't have too much experience with AI, I am currently in a Machine Learning class here at Temple. I'm actually just getting started with TensorFlow now. While my knowledge with this library isn't extensive, it will at least be fresh in my head. In terms of coding language, I have some experience with Python so that wouldn't be too much of an issue.