Project II
CS475
---------------------------------------------------------------
GOAL: Implement a planner that employs search techniques to find
plan(s).

DUE DATE: 5/3/2001

GRADE: 10 points (10% of the final grade) 
---------------------------------------------------------------

DESCRIPTION: You are asked to select a search algorithm and
implement a planner which - when given a planning problem as
input - returns the solutions to the problem if one exists
and return NO otherwise.

A planning problem consists of:
    - an action theory consisting of the set of fluents,
      the set of actions with their effects, and the
      initial state
    - a goal

A solution is a sequence of actions, which transforms the
initial state to the goal state

The format of the input files will be as follows.

;Fluent 
fluent_name_1
fluent_name_2
.....

;Action
name: action_name_1
condition: preconditions of action_name_1
add: positive effects of action_name_1
delete: negative effects of action_name_1

;Action
name: action_name_2
condition: preconditions of action_name_2
add: positive effects of action_name_2
delete: negative effects of action_name_2

....

;Initial
fluent_name_i1
fluent_name_i2
.....

;Goal
fluent_name_g1
fluent_name_g2
.....

A sample data file is given at 

http://www.cs.nmsu.edu/~tson/classes/data_project2


NOTES:

1. You can use any algorithm in the text book.

2. My recommendations:

    - It is good to have a procedure/function that reads the
    input file and stores the information about fluents/actions
    in a data structure. Since the search process will access this
    structure a lot, you might think about a hash table.

    - Forward-chaining seems to be the simplest algorithm for finding
    a plan.  But it is known that it is complete. To have the completeness,
    you might think about using the A* algorithm.

3. Grading:

    - Program that has compiler errors gets 0
    - The planner will be tested with 3-5 domains (that will be
    posted in the class web-site). At the grading time, the TA will
    generate the initial and goal states. Planner, that returns
    correct answers will get 2-3 points for each domain.