is this java interface tactic sound?

I've been thinking about implementing a certain tactic for my code.

This is my setup:

I've got an interface called "Object". Then I've got an interface called "Entity" that extends "Object". From entity then springs countless implementations, like "army", "city", "lemon", etc.

Now, I want to gather all of these Objects into some form of map. Then from that map I want to get the particular implementation of "Object".

My thought out solution for this is as follows:

Object has method :

public Entity getEntity()

All implementations of Object returns null, while Entity returns itself.

Likewise, in entity I'd have:

public Army getArmy()
public City getCity()

That way, I can simply pull an object from the map and get the specific class from it with a series of null checks, like so;

Object o = Objects.getObject(2dCoordinates);
Entity e = o.getEntity();
if (e != null){
Army a = e.getArmy();
if (a != null)
a.armySpecificMethod();
}

All without using "instanceof" and casting, which I hate.

The question is whether there's some unforeseen problem about this? I'd rather learn from someone that knows before refactoring my code and find out for myself.

via Chebli Mohamed