This is student project of AI rat saper defusing bomb on randomly generated field with use of decision tree, genetic algorith, a* algorithm and image recognition with neural networks.
How it works:
- Field is randomly generated in matrix where:
- T - terrain
- B - bomb
- R - rock
- P - puddle
- Rat can not move through rocks, going through puddles is slower than normal terrain
- Genetic algorithm finding the shortest way to move through all bomb squares
- A* algorithm finding best way from one bomb square to another with respect of those rules:
- Rat can go through terrain and bomb squares normally
- Rat moving through puddles slower than through terrain and bomb squares
- Rat can not move through rocks
- When on bomb square, rat judges with image recognition whether its a rock or a mine (if rock, then field changing icon to rock)
- If its a mine, rat judges what to do with it basing on decision tree:
- Take away to sapper grounds (field changing icon to red flag)
- Defuse here (field changing icon to crossed mine)
- Program ends when rat will cover all bomb squares
Presentation: