CSE-571-TeamProject

Team name

Reboot

Team Members

1. Sahit Jain

2. Nabeel Khan

3. Anantha Krishnan Kumar

4. Akshay Ganesa

Topic chosen

Topic 5: Reinforcement Learning agent in Wumpus world

Github Link: https://github.com/nakhan8/CSE-571-TeamProject

Project Structure

Src/: Code for project

Doc/: Report for project

Readme.md: Information about project

Contributions of each team member

Sahit Jain-

1. Noisy logic Agent
2. Learning prerequisites for design
3. Studied and explored about implementation

Anantha Krishnan Kumar-

1. Reinforcement Learning
2. Learning prerequisites for design
3. Studied and explored about implementation

Akshay Ganesa-

1. Reinforcement Learning
2. Learning prerequisites for design
3. Studies and explored about implementation

Nabeel Khan-

1. Exploring and familiarizing with repository
2. Create basic project structure for design
3. Created report for the submission

Instructions to run code from scratch

The Hybrid Logic agents reuses most of the code from "The Hunt The Wumpus AI project was developed at University of Arizona by Clay Morrison ([email protected])" that we got in individual project 3.

Knowledge base implented in: "wumpus_kb.py"

Planning: "wumpus_planner.py"

Noise: "wumpus_agent.py"

Prerequisites: Python Installed

1. Dowload the Project.
2. Extract the project
3. Run the project.
4. Execution command:python environment.py --ngames 150 --niter 40 --gridsize 4 4 --numwumpus 1 --numholes 3 --bullets 3

None of the arguments are compulsory.

ngames indicates are number of games

niter indicates number of iterations

gridsize 4 4 indicates rowsize columnsize

numwumpus indicates of Wumpus in grid

numholes indicate number of holes in the grid

bullets indicate the number of bullets that agent possesses

5. For executing the Hybrid Logic Agent in a noisy action model. Just run "python wumpus.py -y"

Noise: The agent goes Forward 90% of the times And takes either a left or a right turn 5% of the times respectively.