# Vacuum Cleaner Simulation


## Goal

The goal of this project is to build a simulation environment for autonomous vacuum cleaners. The goal of a vacuum cleaner should be

1- To clean as much of the dirt as possible.
2- To go back to the home station before the time is over.
3- To use as little memory as possible for the agent.

## Simulation components

The goal of the simulation is that of finding a good strategy for cleaning the area and finishing in the charging station.


### The map

A simulation happens in a map of the environment. There is an class (Map) that models this environment and keeps track of the dirt, the location of the obstacles and the position of interesting objects:

1- The vacuum cleaner, or agent.
2- The charging station.

The map is essentially a matrix and the agent can move from the current cell to one of its 4 neighbors. If the agent tries to move to a cell occupied by an obstacle, nothing happens (the agent wastes time, however).

The dirt, if any, goes away from a cell once the agent moves over it.


### The agent

The job of cleaning is left to the vacuum cleaner, aka 'the agent'. This is modeled in the Agent class. Each agent can move to the 4 neighboring cells or do nothing. The agent can read the remaining battery life and the remaining time (but, of course, cannot change them). There is a charging stations somewhere and if the agent stops (does nothing) over the charging station, the battery gets charged, one unit per time.

The agent feels the presence of obstacles in the 4 neighboring cells. Also, the agent feels the (noisy) signal from the charging station that models the agent-station distance.

### The charging station

No matter how good a vacuum cleaner can clean the floor. If the vacuum cleaner does not finish its job in the charging station, a human will have to move it there. Since humans are lazy, the vacuum cleaner gets a big penalty for not ending up where it should.


### The score

At the end of the available time the agent gets a score for the cleaning job. The final score is given by:

Score = 0.5 (R + H)

where
- R: the ratio of 'removed dirt' / 'initial dirt'.
- H: 1 if the agent is in the charging station, 0 otherwise.

The score ranges between 0 and 1 and 1 is the best the agent can hope for.