Play the rock paper scissors game against an AI!

We developed a Rock-Paper-Scissors game with Python and MediaPipe, where a human player plays against different AI strategies (random move, 20 % cheated move, move determined using Markov chain). Human moves are detected from live webcam images, AI moves are visualized using icons on the screen. 21 study participants (N = 21) could play as long as they wanted, measuring the engagement time with a stopwatch. A questionnaire was used to collect additional data. It was first supposed that the more intelligent the AI strategy (Markov Chain > 20 % cheat > random move), the higher the engagement time will be. However, the data collected shows the opposite result: the less intelligent and the more "fair" the AI strategy, the longer the participants wanted to play. Additional data was analyzed, before and after playing the game, regarding the confidence of playing Rock-Paper-Scissors, the emotions, and more.

Connect a camera to your machine. It is used to determine the player's move.

You can optionally use a virtualenv or Conda environment.

1. Install Python version 3
2. pip install -r requirements.txt
3. python main.py

You can choose from the following strategies that the AI will follow to determine its move:

• Random (RANDOM)
• Cheat with 20% chance, such that it will win against the player (SOMETIMES_CHEAT)
• Markov chain approach based on the last and current player's move (MARKOV_CHAIN)

Currently, the AI strategy is set at the beginning of the main.py file.