steffen/FIKS_Entwicklungsumgebung
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Monitor and Agent State Debug

Browse Source
This commit is contained in:
steffen-illium
2021-05-14 13:39:24 +02:00
parent 18be35dc19
commit 35ef708b20
2 changed files with 32 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files

30
environments/factory/simple_factory.py
View File

@ -1,5 +1,5 @@
import numpy as np
from environments.factory.base_factory import BaseFactory
from environments.factory.base_factory import BaseFactory, FactoryMonitor
class SimpleFactory(BaseFactory):
@ -14,28 +14,36 @@ class SimpleFactory(BaseFactory):
self.state[-1, x, y] = 1
def reset(self):
super().reset()
state, r, done, _ = super().reset()
dirt_slice = np.zeros((1, *self.state.shape[1:]))
self.state = np.concatenate((self.state, dirt_slice)) # dirt is now the last slice
self.spawn_dirt()
# Always: This should return state, r, done, info
return self.state, r, done, _
def calculate_reward(self, agent_states):
for agent_state in agent_states:
collisions = agent_state.collisions
entities = [self.slice_strings[entity] for entity in collisions]
for entity in entities:
self.monitor.add(f'{entity}_collisions', 1)
print(f't = {self.steps}\tAgent {agent_state.i} has collisions with '
f'{entities}')
if entities:
for entity in entities:
self.monitor.add(f'agent_{agent_state.i}_collision_{entity}', 1)
print(f't = {self.steps}\tAgent {agent_state.i} has collisions with '
f'{entities}')
return 0, {}
if __name__ == '__main__':
import random
factory = SimpleFactory(n_agents=1, max_dirt=8)
random_actions = [random.randint(0, 7) for _ in range(200)]
for action in random_actions:
state, r, done, _ = factory.step(action)
print(f'Factory run done, reward is:\n {r}')
print(f'There have been the following collisions: \n {dict(factory.monitor)}')
monitor_list = list()
for epoch in range(100):
random_actions = [random.randint(0, 7) for _ in range(200)]
state, r, done, _ = factory.reset()
for action in random_actions:
state, r, done, info = factory.step(action)
monitor_list.append(factory.monitor)
print(f'Factory run done, reward is:\n {r}')
print(f'There have been the following collisions: \n {dict(factory.monitor)}')