COMP9414 Arti cial Intelligence
GridWorldEnv Environment User Guide
Term 2, 2025
Version: 1.0
Last Updated: July 2025
Purpose: This guide provides comprehensive documentation for the GridWorldEnv class used in Assignment 2
1 Overview
The GridWorldEnv class implements a grid world environment for reinforcement learn- ing experiments. This guide covers installation, usage, and troubleshooting for the envi- ronment provided in env .py.
2 Environment Speci cations
The environment implements an 11 × 11 grid world with the following characteristics:
{ Grid Size: 11 × 11 cells
{ State Space: Discrete 2D coordinates (x;y) where x;y ∈ {0;1;:::;10}
{ Action Space: 4 discrete actions
{ Goal Position: (10; 10) - bottom-right corner
{ Starting Position: Random (avoiding obstacles and goal)
{ Episode Termination: Only when goal is reached
2.1 Coordinate System
The environment uses a standard 2D coordinate system:
o Origin (0; 0) is at the top-left corner
o X-axis increases downward (rows)
o Y-axis increases rightward (columns)
2.2 Obstacles
The environment contains 10 xed obstacles arranged in two patterns:
L-shaped pattern (top-left area):
o Positions: (2; 2), (2; 3), (2; 4), (3; 2), (4; 2)
Cross pattern (centre):
o Positions: (5; 4), (5; 5), (5; 6), (4; 5), (6; 5)
2.3 Action Space
|
Action
|
Value
|
E ect
|
|
Up
|
0
|
Decrease x by 1
|
|
Down
|
1
|
Increase x by 1
|
|
Left
|
2
|
Decrease y by 1
|
|
Right
|
3
|
Increase y by 1
|
2.4 Reward Structure
|
Event
|
Reward
|
|
Reach Goal
Hit Obstacle
Normal Step
|
+25
-10
-1
|