def __init__(self, args):
Action.__init__(self, args)
(brush, waterBucket, color) = self.args
self.preconditions = [('clear', waterBucket), ('holding', brush),
('paintsColor', brush, color)]
self.positiveEffects = [('clean', brush)]
self.negativeEffects = [('paintsColor', brush, color)]
def __init__(self, args):
Action.__init__(self, args)
(obj, sprayer, color) = self.args
self.preconditions = [('onTable', obj), ('clear', obj),
('paintsColor', sprayer, color),
('holding', sprayer)]
self.positiveEffects = [('colored', obj, color)]
self.negativeEffects = []
def solve_pyplanners(instantiated):
# https://github.mit.edu/caelan/stripstream/blob/c8c6cd1d6bd5e2e8e31cd5603e28a8e0d7bb2cdc/stripstream/algorithms/search/pyplanners.py
import sys
pyplanners_path = get_pyplanners_path()
if pyplanners_path is None:
raise RuntimeError(
'Must clone https://github.com/caelan/pyplanners '
'and set the environment variable {} to its path'.format(
PYPLANNERS_VAR))
if pyplanners_path not in sys.path:
sys.path.append(pyplanners_path)
# TODO: could operate on translated SAS instead
from strips.states import State, PartialState
from strips.operators import Action, Axiom
from strips.utils import default_derived_plan
import pddl
if instantiated is None:
return None, INF
py_actions = []
for action in instantiated.actions:
py_action = Action({'fd_action': action})
py_action.conditions = set(action.precondition)
py_action.effects = set()
for condition, effect in action.del_effects:
assert not condition
py_action.effects.add(effect.negate())
for condition, effect in action.add_effects:
assert not condition
py_action.effects.add(effect)
py_action.cost = action.cost
py_action.test = get_attachment_test(action)
py_actions.append(py_action)
py_axioms = []
for axiom in instantiated.axioms:
py_axiom = Axiom({'fd_axiom_id':
id(axiom)}) # Not hashable for some reason
py_axiom.conditions = set(axiom.condition)
py_axiom.effects = {axiom.effect}
py_axioms.append(py_axiom)
goal = PartialState(instantiated.goal_list)
fluents = {f.positive() for f in goal.conditions}
for py_operator in py_actions + py_axioms:
fluents.update(f.positive() for f in py_operator.conditions)
initial = State(atom for atom in instantiated.task.init
if isinstance(atom, pddl.Atom) and (atom in fluents))
plan, state_space = default_derived_plan(initial, goal, py_actions,
py_axioms)
if plan is None:
return None, INF
actions = [pddl_from_instance(action.fd_action) for action in plan]
return actions, plan.cost
def solve_pyplanners(instantiated,
planner=None,
max_planner_time=DEFAULT_MAX_TIME,
max_cost=INF):
if instantiated is None:
return None, INF
# https://github.mit.edu/caelan/stripstream/blob/c8c6cd1d6bd5e2e8e31cd5603e28a8e0d7bb2cdc/stripstream/algorithms/search/pyplanners.py
pyplanners_path = get_pyplanners_path()
if pyplanners_path is None:
raise RuntimeError(
'Must clone https://github.com/caelan/pyplanners '
'and set the environment variable {} to its path'.format(
PYPLANNERS_VAR))
if pyplanners_path not in sys.path:
sys.path.append(pyplanners_path)
# TODO: could operate on translated SAS instead
from strips.states import State, PartialState
from strips.operators import Action, Axiom
from strips.utils import solve_strips, default_derived_plan
import pddl
default_planner = {
'search': 'eager',
'evaluator': 'greedy',
'heuristic': 'ff',
'successors': 'all',
}
if isinstance(planner, dict):
default_planner.update(planner)
py_actions = []
for action in instantiated.actions:
#action.dump()
py_action = Action({'fd_action': action})
py_action.conditions = set(action.precondition)
py_action.effects = set()
for condition, effect in action.del_effects:
assert not condition
py_action.effects.add(effect.negate())
for condition, effect in action.add_effects:
assert not condition
py_action.effects.add(effect)
py_action.cost = action.cost
py_action.test = get_attachment_test(action)
py_actions.append(py_action)
py_axioms = []
for axiom in instantiated.axioms:
#axiom.dump()
py_axiom = Axiom({'fd_axiom_id':
id(axiom)}) # Not hashable for some reason
py_axiom.conditions = set(axiom.condition)
py_axiom.effects = {axiom.effect}
py_axioms.append(py_axiom)
goal = PartialState(instantiated.goal_list)
fluents = {f.positive() for f in goal.conditions}
for py_operator in py_actions + py_axioms:
fluents.update(f.positive() for f in py_operator.conditions)
initial = State(atom for atom in instantiated.task.init
if isinstance(atom, pddl.Atom) and (atom in fluents))
plan, state_space = solve_strips(initial,
goal,
py_actions,
py_axioms,
max_time=max_planner_time,
max_cost=max_cost,
**default_planner)
if plan is None:
return None, INF
actions = [pddl_from_instance(action.fd_action) for action in plan]
#print(actions)
return actions, plan.cost