Skip to content

Rogue-like AI agent for an adaptation of DreamQuest game.

License

Notifications You must be signed in to change notification settings

div5yesh/ai-explorer

Repository files navigation

CMSC 671 (Fall 2018) Final project

Code for the final project of CMSC 671 course (Fall 2018)

To run the code, use the following command:

$ python play.py --height 10 --width 10 \
  --num-powerups 2 --num-monsters 1  --num-dynamic-monsters 1\
  --initial-strength 100 \
  --save-dir map1/ \
  --verbose --show-map --map-type emoji

If you want to play against a human (having a human player as an agent in addition to the random or your implemented agent), use the --play-against-human flag when calling play.py.

To implement a new agent, create a new agent class inheriting from the BaseAgent class in the agent.py file. You just need to implement the step(...) function. Take a look at how RandomAgent has been implemented. You can also switch to HumanAgent to play by hand and see map printouts.

Requirements

  • Python >= 3.6
  • Numpy
  • Emoji (package: emoji) -- if you want emoji maps

Driver Code: v0.1.2

Refer: https://github.com/erfannoury/cmsc671-fall2018

Contributers:

Anushree Desai, Divyesh Chitroda, Hakju Oh, Shree Hari