documentation.txt

Shortest path finding AI

Solution #1
For solving the AI problem I've chosen an approach closer to human 
thinking, instead of brute-forcing through all possible steps. 
Steps:
- Based on the player’s position, the AI creates a blueprint 
around itself, discovering the current sector's available 
options before choosing one. 
- While making a decision the AI is placed to the appropriate
location, based on the decision.
- A path is built, from its original location to the destination, 
and it is stored in its step history.
- After making a decision, the same process is repeated untill 
all, potentially winning decisions are tested. 
- The winning positions are tracked back.
Problems:
- Circumstantial implementation:
This solution requires a lot more debugging then an a simpler 
approach.
Improvement ideas:
- Instead of forging the steps history, the player could walk 
through the generated path, and making sure it gets there. This 
would validate the path as well.

Solution #2
I tried another approach, a recursive backtracking in-depth search. 
This program is much simpler then the previous one, however it has 
one flaw. It can not determine whether the problem can be solved or 
not. It can only try to solve the maze until a determined depth 
limit.

Solution #3
My 3rd solution would be using the first two step of solution#1, to 
determine how many decisions are needed to solve the maze, and using
this depth to solution#2’s recursion.