Introduction

In this assignment you will implement a simulator environment for an automatic vacuum cleaner robot, a set of different reflex$-$based agent programs, and perform a comparison study for cleaning a single room. Focus on the cleaning phase which starts when the robot is activated and ends when the last dirty square in the room has been cleaned. Someone else will take care of the agent program needed to navigate back to the charging station after the room is clean.

PEAS description of the cleaning phase

Performance Measure: Each action costs $1$ energy unit. The performance is measured as the sum of the energy units used to clean the whole room.

Environment: A room with n$\backslash$times n squares where n$=5.$ Dirt is randomly placed on each square with probability p$=0\mathrm{.}2.$ For simplicity, you can assume that the agent knows the size and the layout of the room $($i$.$e$.,$ it knows n $).$ To start, the agent is placed on a random square.

Actuators: The agent can clean the current square $($action suck$)$ or move to an adjacent square by going north, east, south, or west.

Sensors: Four bumper sensors$,$ one for north, east, south, and west; a dirt sensor reporting dirt in the current square.

Task $1$: Implement a simulation environment $[20$ Points$]$

The simple environment above is not very realistic. Your environment simulator needs to follow the PEAS description from above. It needs to:

Initialize the environment by storing the state of each square $($clean$/$dirty$)$ and making some dirty. $($Help with random numbers and arrays in Python$)$

Keep track of the agent's position.

Call the agent function repeatedly and provide the agent function with the sensor inputs.

React to the agent's actions. E$.$g$,$ by removing dirt from a square or moving the agent around unless there is a wall in the way.

Keep track of the performance measure. That is$,$ track the agent's actions until all dirty squares are clean and count the number of actions it takes the agent to complete the task.