An AI solution for the pacman agent to find paths through his maze world, both to reach a particular location and to collect food efficiently. The project consisited of two phases :
- Search Phase - This phase consists of searching in the maze and doesn't involve ghosts etc. Wrote python codes for searching the maze using Depth-first-search, Breadth-First-search algorithms. Compared their preformance in terms of search time and number of nodes expanded against the uniform cost search algorithm. Then implemented the A* algorithm with manhattan heuristic to effeciently search the maze by cutting down the number of states. Also, explored the corners problem and eating the dots problem. My code was able to search less than 9000 nodes in 15 s and recieved 90% score on the AI UC-Berkley challenge.
- Multi-agents phase - To be updated...
****** The GUI for the code as well as some runtime features were adopted from the UC Berkley Open Course Ware for Artificial Intelligence (CS 188) *******