nuked88/genericgamebot-ai

Generalization of FastAI CNN for playing videogames.

Python / FastAI CNN - Playing Game

This code was used to gather and process data while playing the game Fall Guys.

This code will be modified for permit the use in every game with a minimum edit of the config.yaml file The network was then trained using the FastAI Libary.

THIS IS A BIG WORK IN PROGRESS

Setup

• Edit the config.yaml [SEE THE SECTION BELOW]
• Open the game in a small window ( i suggest something 4:3 like 480 x 480 for now)
• Run CreateData.py
• Start changing the value of the region_capured in the config file until the game is perfectly visible in the black window
• Pressing B will start the screen / key grab. These will be stored in lists until you press H.
• Once the episode ( Round ) ends pressing h will stop the screen / key grab process and all data will be moved to a numpy array.The array will be stored in data/target_data_<nameofthegame>.npy and data/training_data_<nameofthegame>.npy
• Go in Utils folder and run CreateImages.py to generate images that will be splitted in folders corresponding to their actions.The images are stored in images/<nameofthegame>/

Example of a configuration file with explanation

game_name:  Tower_Hero #name of the game that will be used for rename the folder and the model (please use underscore intead of space)
path_game: test_path #not used for now
region_captured:  #where the window of the game is located on the screen
  left: 710 #starting point at the left
  top: 300 #starting point at the top
  extensionOfX: 1110 #how far the window of the game extends on the X axis (starting from the left point)
  extensionOfY: 807 #how far the window of the game extends on the Y axis (starting from the top point)
dummmy_direction: LEFT_KEY #direction where to point when you are doing nothing in the game
commands: #association of -key pressed- : -name of the folder where the image will be stored-
  UP_KEY : 'Nothing'
  LEFT_KEY : 'Left'
  DOWN_KEY : 'Down'
  RIGHT_KEY : 'Right'
  ' ' : 'Fire'

Train

• Run training.py

Run Agents

• Open the game withe the window in the same position of the CreateData.py operation
• Run TrainedAgent.py

OLD INSTRUCTION

Setup

• I used dual monitors with a game playing at a 800 X 600 resolution on one screen.
• You can start data collection by running CreateData.py
• Pressing B will start the screen / key grab. These will be stored in lists until the episode is done.
• Once the episode ( Round ) ends pressing h will stop the screen / key grab process and all data will be moved to a numpy array.
• Then I used a script in util folder called CreateImages.py to put then onto a disk drive in folders corresponding to their actions.

Train

• Use the file called training.py
• Point it at your image directory

Run Agents

• Fully random agent is RandomAgent.py
• Trained Agent is TrainedAgent.py
• You will have to load in the pkl created from training.

Inputs (Observations)

• Uses inputs to the neural network (Observations) of pixes in the game.
• 224 X 224
• Line detection

Contact!

• YouTube Clarity Coders
• Chat with me! Discord