Exploring AI Planning and Search Algorithms in Invert

This thesis looks at how two AI methods, A* search and AI planning, can solve levels in the Invert puzzle game. The game has simple rules, but becomes hard because players must flip tiles to the same colour using a limited number of moves. The A* method uses different ways to guess which moves are the best, while the planning method uses PDDL and tools like Fast Downward, Scorpion, and Delfi. The tests compare how fast the methods are, how much memory they use, and how many steps they take to finish a level. The results show that weighted A* works best on small maps, while clustered mismatch A* helps with medium ones. On large maps, planning tools like Delfi and Fast Downward works better than A*. No methods could solve the biggest maps in time or without using too much memory. This study helps show which AI methods work best for games with limited resources.