Build a Pacman AI-Agent

Table of Contents

• Chapter 1 - My Setup Checklist
• Chapter 2 - Info from Udacity
• Chapter 3 - References

Chapter 1 - My Setup Checklist

Instructions and Setup Environment

• Switch to Miniconda environment source activate aind
• Install dependencies python -m pip install mypy typing
• IntelliJ: File > Project Structure > Project Settings > Project > Project SDK > Python 3.6 (~/miniconda3/bin/python)
• Review possible commands in search/commands.txt:
• Change into the search directory cd search
• Show Help of possible commands python pacman.py -h
• Play Pacman and use arrow keys to control and capture ghosts python pacman.py
• Run GoWest agent on map with no turning required python pacman.py --layout testMaze --pacman GoWestAgent
• Run GoWest agent on map with turning required python pacman.py --layout tinyMaze --pacman GoWestAgent
• Run SearchAgent using tinyMazeSearch Algorithm that plans a path to navigate maze successfully python pacman.py -l tinyMaze -p SearchAgent -a fn=tinyMazeSearch
• Run Program with a possible command: python pacman.py --layout smallClassic --zoom 2
• Run MyPy Linter with mypy search/pacman.py

IntelliJ Debugging Setup

• Create new Python debug configuration script
• Select python.py as the Script
• Add -l tinyMaze -p SearchAgent -a fn=depthFirstSearch -z 0.5 ad the script parameters
• Select Python 3.6 specified interpreter (i.e. Miniconda Python 3.6)
• Ensure that .py type files are being included. Press CMD+, then go to Editor > File Types > Python and add .py in addition to .pyw that may be the only one that already exists. Click OK and wait for IntelliJ to reindex
• Now you can set breakpoints in the search.py file and run Debug

Questions

• Goal: Run Algorithms I create that formulate a plan to navigate the maze

• Question 1: DFS Algorithm using Graph Search and avoid expainding previously visited states: python pacman.py -l tinyMaze -p SearchAgent -a fn=depthFirstSearch -z 0.5

Autograder

• Download autograder from here
• Copied/pasted the following as a minimum to autograde Question 1

test_cases/CONFIG
test_cases/q1/*
autograder.py
grading.py
projectParams.py
searchTestClasses.py
testClasses.py
testParser.py

• Run the following commands to convert code from Python 2 to Python 3

2to3 autograder.py -w
2to3 grading.py -w
2to3 projectParams.py -w
2to3 searchTestClasses.py -w
2to3 testClasses.py -w
2to3 testParser.py -w

• Delete all the .bak files that are generated

• Show available autograder commands: python autograder.py -h

• Run the autograder against Question 1 python autograder.py --question=q1

Hints

• Try solving DFS Algorithm first
• Use Stack, Queue and PriorityQueue types in util.py

Objective

• - Score of 20

Files Used

• search.py - Search Algorithms (i.e. BFS, DFS, A*)

• searchAgent.py - AI-agent

• pacman.py - runs Pacman game (with GameState type)

• game.py - logic describing types: AgentState, Agent, Direction, Grid

• util.py - data structures for implementing Search Algorithms

Chapter 2 - Info from Udacity

• Goal: Help Pac-Man navigate his world in the most efficient way to find food and other objects by implementing BFS, DFS, and A* Search.

Chapter 3 - References

• Pacman search code https://d17h27t6h515a5.cloudfront.net/topher/2017/January/587da420_pacman/pacman.zip
• Pacman problem set instructions https://inst.eecs.berkeley.edu/~cs188/fa10/projects/search/search.html
• Instructions https://inst.eecs.berkeley.edu/~cs188/fa10/projects/search/search.html