def taxid():
env = init_taxi_d(seed=1234)
target = list(env.decode(env.s))
print(target)
goalspec = 'F P_[PI]['+str(3)+',none,==]'
keys = ['L', 'PI', 'DI']
actions = [0, 1, 2, 3, 5]
root = goalspec2BT(goalspec, planner=None)
behaviour_tree = BehaviourTree(root)
child = behaviour_tree.root
planner = GenRecPropTaxi(
env, keys, child.name, dict(), actions=actions,
max_trace=5, seed=123)
child.setup(0, planner, True, 5)
# 4,3,4,3
# print(child.goalspec, child.planner.goalspec, child.planner.env)
for i in range(5):
behaviour_tree.tick(
pre_tick_handler=reset_env_d(env)
)
print(i, behaviour_tree.root.status)
child.train = False
for i in range(1):
behaviour_tree.tick(
pre_tick_handler=reset_env_d(env)
)
print(i, behaviour_tree.root.status)
def setUp(self):
env = init_taxi_s(seed=1234)
# target = list(env.decode(env.s))
goalspec = 'F P_[PI]['+str(4)+',none,==]'
keys = ['L', 'PI', 'DI']
actions = [0, 1, 2, 3, 4, 5]
root = goalspec2BT(goalspec, planner=None)
self.behaviour_tree = BehaviourTree(root)
child = self.behaviour_tree.root
planner = GenRecPropTaxi(
env, keys, child.name, dict(), actions=actions,
max_trace=5, seed=123)
child.setup(0, planner, True, 5)
print(child.goalspec, child.planner.goalspec, child.planner.env)
for i in range(5):
self.behaviour_tree.tick(
pre_tick_handler=reset_env_s(env)
)
# child.setup(0, planner, True, 40)
child.train = False
print(child, child.name, child.train)
for i in range(1):
self.behaviour_tree.tick(
pre_tick_handler=reset_env_s(env)
)
print('inference', self.behaviour_tree.root.status)
def taxi():
env = init_taxi(seed=1234)
target = list(env.decode(env.s))
print(target)
goalspec = '((((F(P_[L]['+give_loc(target[2])+',none,==])) U (F(P_[PI]['+str(4)+',none,==]))) U (F(P_[L]['+give_loc(target[3])+',none,==]))) U (F(P_[PI]['+str(target[3])+',none,==])))' # noqa: E501
# goalspec = 'F P_[L]['+give_loc(target[2])+',none,==] U F P_[PI]['+str(4)+',none,==]' # noqa: E501
keys = ['L', 'PI', 'DI']
# actions = [[0, 1, 2, 3], [4], [0, 1, 2, 3], [5]]
actions = [0, 1, 2, 3, 4, 5]
root = goalspec2BT(goalspec, planner=None)
print('root', root)
behaviour_tree = BehaviourTree(root)
# display_bt(behaviour_tree)
epoch = [80, 50, 80, 50]
j = 0
for child in behaviour_tree.root.children:
print('children', child, child.name, child.id)
planner = GenRecPropTaxi(
env, keys, child.name, dict(), actions=actions,
max_trace=40, seed=1)
child.setup(0, planner, True, epoch[j])
j += 1
# planner.env = env
# print(child.goalspec, child.planner.goalspec, child.planner.env)
# print('rootname', behaviour_tree.root.name)
# behaviour_tree.root.remove_child_by_id(id)
# display_bt(behaviour_tree)
for i in range(200):
behaviour_tree.tick(
pre_tick_handler=reset_env(env)
)
print('Training', behaviour_tree.root.status)
for child in behaviour_tree.root.children:
# child.setup(0, planner, True, 20)
child.train = False
# print(child, child.name, child.train)
for i in range(2):
behaviour_tree.tick(
pre_tick_handler=reset_env(env)
)
print('Inference', behaviour_tree.root.status)
def setUp(self):
env = init_taxi(seed=1234)
target = list(env.decode(env.s))
goalspec = 'F(P_[L]['+give_loc(target[2])+',none,==])'
keys = ['L', 'PI', 'DI']
actions = [0, 1, 2, 3]
root = goalspec2BT(goalspec, planner=None)
self.behaviour_tree = BehaviourTree(root)
child = self.behaviour_tree.root
planner = GenRecPropTaxi(
env, keys, None, dict(), actions=actions,
max_trace=40, seed=1234)
child.setup(0, planner, True, 50)
print(child.goalspec, child.planner.goalspec, child.planner.env)
for i in range(50):
self.behaviour_tree.tick(
pre_tick_handler=reset_env(env)
)
def setUp(self):
env = init_taxi(seed=1234)
target = list(env.decode(env.s))
print(target)
goalspec = '((((F(P_[L]['+give_loc(target[2])+',none,==])) U (F(P_[PI]['+str(4)+',none,==]))) U (F(P_[L]['+give_loc(target[3])+',none,==]))) U (F(P_[PI]['+str(target[3])+',none,==])))' # noqa: E501
keys = ['L', 'PI', 'DI']
actions = [0, 1, 2, 3, 4, 5]
root = goalspec2BT(goalspec, planner=None)
# print('root', root)
self.behaviour_tree = BehaviourTree(root)
epoch = [80, 50, 80, 50]
j = 0
for child in self.behaviour_tree.root.children:
# print('children', child, child.name, child.id)
planner = GenRecPropTaxi(
env, keys, child.name, dict(), actions=actions,
max_trace=40, seed=1)
child.setup(0, planner, True, epoch[j])
j += 1
# Training
for i in range(200):
self.behaviour_tree.tick(
pre_tick_handler=reset_env(env)
)
print('Training', self.behaviour_tree.root.status)
# Inference
for child in self.behaviour_tree.root.children:
child.train = False
# print(child, child.name, child.train)
for i in range(2):
self.behaviour_tree.tick(
pre_tick_handler=reset_env(env)
)
print('inference', self.behaviour_tree.root.status)