def __init__(self, cargos, planes, airports, initial: FluentState, goal: list):

"""

:param cargos: list of str

cargos in the problem

:param planes: list of str

planes in the problem

:param airports: list of str

airports in the problem

:param initial: FluentState object

positive and negative literal fluents (as expr) describing initial state

:param goal: list of expr

literal fluents required for goal test

"""

self.state_map = initial.pos + initial.neg

self.initial_state_TF = encode_state(initial, self.state_map)

Problem.__init__(self, self.initial_state_TF, goal=goal)

self.cargos = cargos

self.planes = planes

self.airports = airports

self.actions_list = self.get_actions()

def get_actions(self):

"""

This method creates concrete actions (no variables) for all actions in the problem

domain action schema and turns them into complete Action objects as defined in the

aimacode.planning module. It is computationally expensive to call this method directly;

however, it is called in the constructor and the results cached in the `actions_list` property.

Returns:

----------

list<Action>

list of Action objects

"""

# TODO create concrete Action objects based on the domain action schema for: Load, Unload, and Fly

# concrete actions definition: specific literal action that does not include variables as with the schema

# for example, the action schema 'Load(c, p, a)' can represent the concrete actions 'Load(C1, P1, SFO)'

# or 'Load(C2, P2, JFK)'. The actions for the planning problem must be concrete because the problems in

# forward search and Planning Graphs must use Propositional Logic

def load_actions():

"""Create all concrete Load actions and return a list

:return: list of Action objects

"""

loads = []

# create all load ground actions from the domain Load action

for a in self.airports:

for p in self.planes:

for c in self.cargos:

precond_pos = [

expr("At({}, {})".format(c, a)), # cargo at airport

expr("At({}, {})".format(p, a)) # plane at airport

]

precond_neg = []

effect_add = [expr("In({}, {})".format(c, p))] # In Plane

effect_rem = [expr("At({}, {})".format(c, a))] # cargo NOT At airport

load = Action(expr("Load({}, {}, {})".format(c, p, a)),

[precond_pos, precond_neg],

[effect_add, effect_rem])

loads.append(load)

return loads

def unload_actions():

"""Create all concrete Unload actions and return a list

:return: list of Action objects

"""

unloads = []

# create all Unload ground actions from the domain Unload action

for a in self.airports:

for p in self.planes:

for c in self.cargos:

precond_pos = [

expr("In({}, {})".format(c, p)), # cargo in plane

expr("At({}, {})".format(p, a)) # plane at airport

]

precond_neg = []

effect_add = [expr("At({}, {})".format(c, a))] # Cargo at Airport

effect_rem = [expr("In({}, {})".format(c, p))] # Cargo NOT in plane

unload = Action(expr("Unload({}, {}, {})".format(c, p, a)),

[precond_pos, precond_neg],

[effect_add, effect_rem])

unloads.append(unload)

return unloads

def fly_actions():

"""Create all concrete Fly actions and return a list

:return: list of Action objects

"""

flys = []

for fr in self.airports:

for to in self.airports:

if fr != to:

for p in self.planes:

precond_pos = [expr("At({}, {})".format(p, fr)),

]

precond_neg = []

effect_add = [expr("At({}, {})".format(p, to))]

effect_rem = [expr("At({}, {})".format(p, fr))]

fly = Action(expr("Fly({}, {}, {})".format(p, fr, to)),

[precond_pos, precond_neg],

[effect_add, effect_rem])

flys.append(fly)

return flys

return load_actions() + unload_actions() + fly_actions()

def actions(self, state: str) -> list:

""" Return the actions that can be executed in the given state.

:param state: str

state represented as T/F string of mapped fluents (state variables)

e.g. 'FTTTFF'

:return: list of Action objects

"""

possible_actions = []

kb = PropKB()

decoded = decode_state(state, self.state_map)

# if we feed decoded.sentence() in, the clauses will also

# include all the negatives like At(C1, SFO) but also ~At(C1, JFK)

# approach adapted from example_have_cake.py

kb.tell(decoded.pos_sentence())

for action in self.actions_list:

is_possible = True

# loop through the action's POSITIVE precondition clauses

for pos_clause in action.precond_pos:

if pos_clause not in kb.clauses:

is_possible = False

# loop through the action's NEGATIVE precondition clauses

for neg_clause in action.precond_neg:

if neg_clause in kb.clauses:

is_possible = False

if is_possible:

possible_actions.append(action)

return possible_actions

def result(self, state: str, action: Action):

""" Return the state that results from executing the given

action in the given state. The action must be one of

self.actions(state).

:param state: state entering node

:param action: Action applied

:return: resulting state after action

"""

# approach adapted from example_have_cake.py

new_state = FluentState([], [])

old_state = decode_state(state, self.state_map)

for fluent in old_state.pos:

# if the action did not remove old positive fluent, keep it

if fluent not in action.effect_rem:

new_state.pos.append(fluent)

for fluent in action.effect_add:

# if we haven't added the resulting fluent to new state, add it

if fluent not in new_state.pos:

new_state.pos.append(fluent)

for fluent in old_state.neg:

# if the old negative fluent hasn't been added, keep it

if fluent not in action.effect_add:

new_state.neg.append(fluent)

for fluent in action.effect_rem:

# if the fluent being removed is not already in the new stat's neg list, add

if fluent not in new_state.neg:

new_state.neg.append(fluent)

return encode_state(new_state, self.state_map)

def goal_test(self, state: str) -> bool:

""" Test the state to see if goal is reached

:param state: str representing state

:return: bool

"""

kb = PropKB()

kb.tell(decode_state(state, self.state_map).pos_sentence())

for clause in self.goal:

if clause not in kb.clauses:

return False

return True

def h_1(self, node: Node):

# note that this is not a true heuristic

h_const = 1

return h_const

@lru_cache(maxsize=8192)

def h_pg_levelsum(self, node: Node):

"""This heuristic uses a planning graph representation of the problem

state space to estimate the sum of all actions that must be carried

out from the current state in order to satisfy each individual goal

condition.

"""

pg = PlanningGraph(self, node.state)

pg_levelsum = pg.h_levelsum()

return pg_levelsum

@lru_cache(maxsize=8192)

def h_ignore_preconditions(self, node: Node):

"""This heuristic estimates the minimum number of actions that must be

carried out from the current state in order to satisfy all of the goal

conditions by ignoring the preconditions required for an action to be

executed.

"""

kb = PropKB()

decoded = decode_state(node.state, self.state_map)

kb.tell(decoded.pos_sentence())

current_clauses = kb.clauses

# Russell-Norvig 382: "Almost implies the number of steps required to solve a relaxed problem

# is the number of unsatisfied goals"

# following gets the number of goal clauses not satisfied,

# works for independent goals

# ref: https://ai-nd.slack.com/archives/C3TPR3RCG/p1502854124000011

unsatified_goals = [goal_clause for goal_clause in self.goal if goal_clause not in current_clauses]

cargos = ['C1', 'C2']

planes = ['P1', 'P2']

airports = ['JFK', 'SFO']

pos = [expr('At(C1, SFO)'),

expr('At(C2, JFK)'),

expr('At(P1, SFO)'),

expr('At(P2, JFK)'),

]

neg = [expr('At(C2, SFO)'),

expr('In(C2, P1)'),

expr('In(C2, P2)'),

expr('At(C1, JFK)'),

expr('In(C1, P1)'),

expr('In(C1, P2)'),

expr('At(P1, JFK)'),

expr('At(P2, SFO)'),

]

init = FluentState(pos, neg)

goal = [expr('At(C1, JFK)'),

expr('At(C2, SFO)'),

]

cargos = ['C1', 'C2', 'C3']

planes = ['P1', 'P2', 'P3']

airports = ['JFK', 'SFO', 'ATL']

pos = [expr('At(C1, SFO)'),

expr('At(C2, JFK)'),

expr('At(C3, ATL)'),

expr('At(P1, SFO)'),

expr('At(P2, JFK)'),

expr('At(P3, ATL)')

]

neg = [# NOT states of C1

expr('At(C1, JFK)'),

expr('At(C1, ATL)'),

expr('In(C1, P1)'),

expr('In(C1, P2)'),

expr('In(C1, P3)'),

# NOT states of C2

expr('At(C2, SFO)'),

expr('At(C2, ATL)'),

expr('In(C2, P1)'),

expr('In(C2, P2)'),

expr('In(C2, P3)'),

# NOT states of C3

expr('At(C3, JFK)'),

expr('At(C3, SFO)'),

expr('In(C3, P1)'),

expr('In(C3, P2)'),

expr('In(C3, P3)'),

# NOT states for P1

expr('At(P1, JFK)'),

expr('At(P1, ATL)'),

# NOT states for P2

expr('At(P2, SFO)'),

expr('At(P2, ATL)'),

# NOT states for P3

expr('At(P3, JFK)'),

expr('At(P3, SFO)')

]

init = FluentState(pos, neg)

# Goal(At(C1, JFK) ∧ At(C2, SFO) ∧ At(C3, SFO))

goal = [expr('At(C1, JFK)'),

expr('At(C2, SFO)'),

expr('At(C3, SFO)')

]

cargos = ['C1', 'C2', 'C3','C4']

planes = ['P1', 'P2']

airports = ['SFO','JFK','ATL','ORD']

pos = [expr('At(C1, SFO)'),

expr('At(C2, JFK)'),

expr('At(C3, ATL)'),

expr('At(C4, ORD)'),

expr('At(P1, SFO)'),

expr('At(P2, JFK)')

]

neg = [# NOT states of C1

expr('At(C1, JFK)'),

expr('At(C1, ATL)'),

expr('At(C1, ORD)'),

expr('In(C1, P1)'),

expr('In(C1, P2)'),

# NOT states of C2

expr('At(C2, SFO)'),

expr('At(C2, ATL)'),

expr('At(C2, ORD)'),

expr('In(C2, P1)'),

expr('In(C2, P2)'),

# NOT states of C3

expr('At(C3, SFO)'),

expr('At(C3, JFK)'),

expr('At(C3, ORD)'),

expr('In(C3, P1)'),

expr('In(C3, P2)'),

# NOT states of C4

expr('At(C4, SFO)'),

expr('At(C4, JFK)'),

expr('At(C4, ATL)'),

expr('In(C4, P1)'),

expr('In(C4, P2)'),

# NOT states for P1

expr('At(P1, JFK)'),

expr('At(P1, ATL)'),

expr('At(P1, ORD)'),

# NOT states for P2

expr('At(P2, SFO)'),

expr('At(P2, ATL)'),

expr('At(P2, ORD)')

]

init = FluentState(pos, neg)

# Goal(At(C1, JFK) ∧ At(C3, JFK) ∧ At(C2, SFO) ∧ At(C4, SFO))

goal = [expr('At(C1, JFK)'),

expr('At(C3, JFK)'),

expr('At(C2, SFO)'),

expr('At(C4, SFO)')

]