Summary: The TetrisAI program was developed to play Tetris 99 on the Nintendo Switch. It utilized computer vision to analyze the game state, a depth-first search algorithm to determine optimal block placement, and a microcontroller to send button presses to the Switch. The algorithm achieved top 15 rankings consistently and even first place occasionally. The project started with a simple idea but evolved into a complex system involving vision algorithms and USB communication. Challenges such as dealing with gameplay effects like sparkles were creatively overcome. The implementation involved Jeff’s eyes for vision, brain for decision-making, and hands for executing moves. The team explored various strategies to optimize gameplay, ultimately achieving success in playing autonomously.
https://bpinzone.github.io/TetrisAI/