Step 01: Implement a simulation environment.
Must haves : environment must be a 2D grid of size M X N.
Environment must be initialized with some dirt randomly on X% of grids.
One grid can have more dirt than another grid.
Must have a performance measure.
Step 02: Implement a randomized agent.
Think of this as a malfunctioning vacuum cleaner.
Randomized agent will move up, down, left, right or suck irrespective of what it percepts.
Step 03: Implement a reflex agent.
A agent that selects actions on the basis of the current percept.
A reflex agent will move up, down, left, right or suck depeding on the current precept.
Step 04: Implement a model based reflex agent.
A agent that selects actions on the basis of the percept history.
A model based reflex agent will move up, down, left, right or suck depending on the precept history.
Step 05: Run simulations.
Report average performance of 100 runs, for all three types of agents on grid sizes (5X5), (10 X 10), (100 X 100).