-
Notifications
You must be signed in to change notification settings - Fork 0
/
EnvironmentManager.py
174 lines (141 loc) · 5.83 KB
/
EnvironmentManager.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
import os
import uuid
from collections import defaultdict
import gym
from PIL import Image
class EnvironmentManager(object):
"""
Environment Manager
Control the environments that will be handled by the server.
"""
TRAINING_DIRECTORY = "monitor/{}/simulation"
RENDER_DIRECTORY = "monitor/{}/rendered/"
def __init__(self):
self.envs = {}
self.render_counter = defaultdict(int)
self.id_len = 8
def create(self, env_id):
"""
Create a new environment.
:param env_id: Valid gym environment.
:return: Instance ID
"""
env = gym.make(env_id)
instance_id = str(uuid.uuid4().hex)[:self.id_len]
self.envs[instance_id] = env
return instance_id
def reset(self, instance_id, render=False):
"""
Reset state of a given instance id.
:param instance_id:
:param render: render result of this operation.
:return: Array of observations.
"""
env = self.envs[instance_id]
initial_state = {'observation': env.observation_space.to_jsonable(env.reset())}
try:
if render:
self.render(instance_id)
initial_state['render'] = 'successfully rendered: {}.png'.format(
self.render_counter[instance_id])
except Exception as e:
initial_state['render'] = 'Error: {}'.format(e.message)
return initial_state
def step(self, instance_id, action, render=False):
"""
Execute action given by agent on the given environment
:param instance_id: instance id of the environment
:param action: action
:param render: render result of this operation.
:return: Dictionary with observation, reward, done flag and info.
"""
env = self.envs[instance_id]
action_from_json = int(env.action_space.from_jsonable(action))
observation, reward, done, info = env.step(action_from_json)
environment_response = {
'observation': env.observation_space.to_jsonable(observation),
'reward': reward,
'done': done,
'info': info
}
try:
if render:
self.render(instance_id)
environment_response['render'] = 'successfully rendered: {}.png'.format(
self.render_counter[instance_id])
except Exception as e:
environment_response['render'] = 'Error: {}'.format(e.message)
return environment_response
def monitor_start(self, instance_id, force, resume):
"""
Start Monitoring.
The directory will be the same as the instace_id.
The video will be available on the port server:8000/instance_id
:param instance_id: instance id of the desired environment.
:param force: Clear out existing training data from this directory (by deleting every file prefixed with "openaigym.").
:param resume: Retain the training data already in this directory, which will be merged with our new data.
:return:
"""
env = self.envs[instance_id]
directory = self.TRAINING_DIRECTORY.format(instance_id)
env.monitor.start(directory, force=force, resume=resume)
def monitor_close(self, instance_id):
"""
Stop Monitor
:param instance_id: instance id.
:return: None.
"""
env = self.envs[instance_id]
env.monitor.close()
def upload(self, instance_id, algorithm_id, writeup, api_key, ignore_open_monitors):
"""
Upload training information created with monitor.
:param instance_id: Id of the environment that was trained.
:param algorithm_id: An arbitrary string indicating the paricular version of the algorithm
(including choices of parameters) you are running.
:param writeup: A Gist URL (of the form https://gist.github.com/<user>/<id>)
containing your writeup for this evaluation.
:param api_key: Your OpenAI API key. Can also be provided as an environment variable (OPENAI_GYM_API_KEY).
:param ignore_open_monitors: Ignore open monitors when uploading.
:return:
"""
directory = self.TRAINING_DIRECTORY.format(instance_id)
gym.upload(directory, algorithm_id, writeup, api_key,
ignore_open_monitors)
def info(self, instance_id):
"""
Expose useful information, such as: action_space, and observation_space.
:param instance_id:
:return:
"""
env = self.envs[instance_id]
return {
'action_space': str(env.action_space),
'observation_space': {
'shape': env.observation_space.shape,
'low': env.observation_space.low.tolist(),
'high': env.observation_space.high.tolist()
}
}
def render(self, instance_id):
"""
Take a screenshot of the simulation.
:param instance_id: id of the environment.
:return:
"""
env = self.envs[instance_id]
if 'rgb_array' in env.metadata['render.modes']:
render_mode = 'rgb_array'
elif 'ansi' in env.metadata['render.modes']:
render_mode = 'ansi'
else:
render_mode = None
if render_mode is None:
raise Exception("Environment does not support 'rgb_array' or 'ansi' renders.")
directory = self.RENDER_DIRECTORY.format(instance_id)
if not os.path.exists(directory):
os.makedirs(directory)
img = Image.fromarray(env.render(render_mode))
img.save(directory + "{}.png".format(self.render_counter[instance_id]))
img.save(directory + "latest.png".format(self.render_counter[instance_id]))
self.render_counter[instance_id] += 1