This is a simple attempt at building an AI capable of playing and improving at Tetris. It uses the classic Genetic Algorithm and Simulated Annealing to learn and improve.
Many features were considered when making this AI and the following were tried and tested:
- Aggregate Height
- Max Height
- Number Of Patches
- Sum Of Depth Of Holes
- Average eight
- Number Of Blocks Above Holes
- Number Of Rows Cleared
- Unevenness
- Number Of Holes
- Number Of Well-Blocks
The final weights obtained in the test runs that we conducted cleared over 4 million lines, with an average of approximately 500,000.
- Suyash Shekhar(@sushinoya)
- Lu Lechuan(@LuLechuan)
- Pan Haozhe(@Haozhe321)
- Shi Tian(@shitian007)
- Kaiser Tan(@kaisertanqr)