Skip to content

Latest commit

 

History

History
44 lines (34 loc) · 3.73 KB

README.md

File metadata and controls

44 lines (34 loc) · 3.73 KB

PacMan With Search Algorithms

In this project Search algorithms and related functions have been developed in line with shared codes [1]. The project was explained with this report. Pacman agent food in line with the project depth first search (DFS), breadth first search (BFS), uniform cost search (UCS) and A * search algorithms are used. All of the auxiliary codes for the project are written in Python 2.7. New in Python 3 Python 3 system in the coding phase, as the code structure was changed with the regulations Sub-version 2.7 is used instead.

The project is used to run using the general pattern given in gameState.py on the terminal. Of used mold its general definition is as in the commands.txt file.

The following search algorithms have been implemented and the scenarios specified has been applied. The implemented algorithms are as follows; I.DFS II. BFS III. UCS IV. A *

The Python files mentioned in terms of content and their explanations are as follows.
pacman.py - Pacman is the main file where the game is executed. Allows creation of Pacman for GameState type. game.py - Pacman forms the logic part of the game, ie the basis. util.py - Contains the data structures required for search algorithms. graphicsDisplay.py - Provides game graphics. graphicsUtils.py - Allows rendering of graphics. textDisplay.py - Provides ASCII graphics for Pacman. ghostAgents.py - Includes agents that control ghosts. keyboardAgents.py - Provides keyboard controls. layout.py - allows to read layout files and store their contents.

The contents of search.py and searchAgents.py used for the search are as follows.
search.py- There are four different algorithms in the Search Problem class (depthFirstSearch, breadthFirstSearch, uniformCostSearch, aStarSearch) searchAgents.py- It includes the goal and instinctual functions of the Pacman agent. (SearchAgent, CornersProblem, cornersHeuristic, AStarCornersAgent,FoodSearchProblem, foodHeuristic,ClosestDotSearchAgent,AnyFoodSearchProblem)

Result

alt text

Continuous neighbor using Pacman stack because DFS is tested in small search space continues by selecting the node. For example, it moves towards the west, choosing its neighbors continuously. More Then it moves south, which is its neighbor. Searches and a cell using the BFS queue continues to search with distance. BFS is found to be better when searching in a small area. BFS for all time guarantees the optimum solution.

alt text
As a result of the comparison, BFS performed better. BFS to give optimal results This is a disadvantage as the memory requirement grows exponentially with depth.

alt text
Since BFS searches on a width basis in the problem of searching for food in the complex map, When searching cells, 16688 node has expanded to search process. But optimum has reached the conclusion. On the other hand, DFS used less cost but could not reach the optimum result.
alt text
BFS ile optimum yola ulaşılmış ancak arama maliyeti oldukça fazla olmuştur. Bu nedenle içgüdüsel yöntem kullanılmış büyük harita üzerinde manhattan ve euclidean uzaklık formülleri kullanarak sonuçlar alınmıştır.
alt text
In line with the scenarios specified, finding the shortest path that touches four corners was implemented using A *.

https://inst.eecs.berkeley.edu/~cs188/sp10/projects/search/ [1].