Genetric AI Tetris

Michael Germano August, 2016 Written in Python3.5

Python AI that can continually clear lines in the arcade game Tetris Built using the tetris terminal game from Eric Pai

Requirements

• pyinput
• numpy

Project Breakdown

• ai.py:
  • analyzes gameboard, calculating all possible moves and subsequent moves based off of the current and next game piece.
  • calculates the score of each possible move based off of
    • lines formed
    • aggregate height
    • the number of holes
    • overall bumpiness
  • executes the best move
• runAI.py:
  • script to start the game and AI
• scoreConstants.py:
  • constants used to calculate the each move's score. These were found using the genetic algorithm.
• geneticAlg:
  • Algorithm that continually improves a population of AI. Each organism in the population contains a list of 4 scoring constants. These constants are intially found randomly using a normal distribution. Every round, the best 50% of the AI breed randomly. Their offspring replace the worst 25% and the algorithm continues on.
• tetris dir:
  • terminal tetris game created by Eric Pai