This game is generated using genetic algorithm NEAT (Neuroevolution of augmenting topologies). The input given to the neural networks are the Current Speedof game, the Horizontal distance and the Verticle distance of dino from obstacles. The dino can jump through obstacles as well as duck through some of them (Flying Dino).
The velocity of the upcoming obstacles keeps on increasing to a certain point. There are two modes for playing the game, the game played by human and game played by machine.
The highest score of both modes are saved and if user crosses these it changes.
- Pygame for making the game.
- NEAT-python for algorithm.
The game starts with asking the user for the mode as shown below.
In the automatic mode the machine starts playing the game
starting with a population of 100.
In the manual mode the user starts playing by pressing SPACE to jump and
DOWN-ARROW-KEY to duck.
1. Youtube video 1 : https://www.youtube.com/playlist?list=PLRqwX-V7Uu6Yd3975YwxrR0x40XGJ_KGO
2. Youtube video 2 : https://www.youtube.com/watch?v=9zfeTw-uFCw&list=PLRqwX-V7Uu6bJM3VgzjNV5YxVxUwzALHV
3. Youtube video 3 : https://www.youtube.com/watch?v=1xkykD5Olok
4. Article 1 : https://towardsdatascience.com/neat-an-awesome-approach-to-neuroevolution-3eca5cc7930f
5. Must read documents are available inside docs folder.
1.5 days. (Half day for that damn duck!)