def get_bt_from_root(self):
"""
Returns bt string (actually a list) from py tree root
by cleaning the ascii tree from py trees
Not complete or beautiful by any means but works for many trees
"""
string = pt.display.ascii_tree(self.root)
print(string)
string = string.replace('[o] ', '')
string = string.replace('\t', '')
string = string.replace('-->', '')
string = string.replace('Fallback', 'f(')
string = string.replace('Sequence', 's(')
bt = string.split('\n')
bt = bt[:-1] #Remove empty element because of final newline
prev_leading_spaces = 999999
for i in range(len(bt) - 1, -1, -1):
leading_spaces = len(bt[i]) - len(bt[i].lstrip(' '))
bt[i] = bt[i].lstrip(' ')
if leading_spaces > prev_leading_spaces:
for _ in range(
round((leading_spaces - prev_leading_spaces) / 4)):
bt.insert(i + 1, ')')
prev_leading_spaces = leading_spaces
bt_obj = behavior_tree.BT(bt)
bt_obj.close()
return bt_obj.bt
def mutate_gene(genome, p_add, p_delete):
"""
Mutate only a single gene.
"""
if p_add < 0 or p_delete < 0:
raise Exception("Mutation parameters must not be negative.")
if p_add + p_delete > 1:
raise Exception(
"Sum of the mutation probabilities must be less than 1.")
mutated_individual = behavior_tree.BT([])
max_attempts = 100
attempts = 0
while (not mutated_individual.is_valid()
or mutated_individual.bt == genome) and attempts < max_attempts:
mutated_individual.set(genome)
index = random.randint(0, len(genome) - 1)
mutation = random.random()
# Delete node
if mutation < p_delete:
mutated_individual.delete_node(index)
# Add node
elif mutation < p_delete + p_add:
mutated_individual.add_node(index)
# Mutate node
else:
mutated_individual.change_node(index)
#Close bt accordingly to the change
mutated_individual.close()
attempts += 1
if attempts >= max_attempts and (not mutated_individual.is_valid()
or mutated_individual.bt == genome):
mutated_individual = behavior_tree.BT([])
return mutated_individual.bt
def crossover_genome(genome1, genome2):
"""
Do crossover between genomes at random points
"""
bt1 = behavior_tree.BT(genome1)
bt2 = behavior_tree.BT(genome2)
offspring1 = behavior_tree.BT([])
offspring2 = behavior_tree.BT([])
if bt1.is_valid() and bt2.is_valid():
max_attempts = 100
attempts = 0
found = False
while not found and attempts < max_attempts:
offspring1.set(bt1.bt)
offspring2.set(bt2.bt)
cop1 = -1
cop2 = -1
if len(genome1) == 1:
cop1 = 0 #Change whole tree
else:
while not offspring1.is_subtree(cop1):
cop1 = random.randint(1, len(genome1) - 1)
if len(genome2) == 1:
cop2 = 0 #Change whole tree
else:
while not offspring2.is_subtree(cop2):
cop2 = random.randint(1, len(genome2) - 1)
offspring1.swap_subtrees(offspring2, cop1, cop2)
attempts += 1
if offspring1.is_valid() and offspring2.is_valid():
found = True
if not found:
offspring1.set([])
offspring2.set([])
return offspring1.bt, offspring2.bt
def __init__(self, string, behaviors, state_machine=None, root=None):
if root is not None:
self.root = root
string = self.get_bt_from_root()
self.bt = behavior_tree.BT(string)
self.depth = self.bt.depth()
self.length = self.bt.length()
self.state_machine = state_machine
self.behaviors = behaviors
if root is None:
self.root, has_children = behaviors.get_node_from_string(
string[0], state_machine)
string.pop(0)
else:
has_children = False
super().__init__(root=self.root)
if has_children:
self.create_from_string(string, self.root)
def random_genome(length):
"""
Returns a random genome
"""
bt = behavior_tree.BT([])
return bt.random(length)