-
Notifications
You must be signed in to change notification settings - Fork 12
/
generate_demonstration_data.py
199 lines (168 loc) · 7 KB
/
generate_demonstration_data.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
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
import gym, os
import numpy as np
import argparse
import random
import pandas as pd
import sys
import torch
from gym import wrappers
import random
import torch.nn.functional as F
import torch.nn as nn
import torch as th
from dril.a2c_ppo_acktr.envs import make_vec_envs
from dril.a2c_ppo_acktr.model import Policy
from dril.a2c_ppo_acktr.arguments import get_args
import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
args = get_args()
args.recurrent_policy = False
args.load_expert = True
os.system(f'mkdir -p {args.demo_data_dir}')
os.system(f'mkdir -p {args.demo_data_dir}/tmp/gym')
sys.path.insert(1,os.path.join(args.rl_baseline_zoo_dir, 'utils'))
from utils import get_saved_hyperparams
#device = torch.device("cpu")
device = torch.device("cuda:0" if args.cuda else "cpu")
print(f'device: {device}')
seed = args.seed
print(f'seed: {seed}')
if args.env_name in ['highway-v0']:
import highway_env
from rl_agents.agents.common.factory import agent_factory
env = make_vec_envs(args.env_name, seed, 1, 0.99, f'{args.emo_data_dir}/tmp/gym', device,\
True, stats_path=stats_path, hyperparams=hyperparams, time=time,
atari_max_steps=args.atari_max_steps)
# Make agent
agent_config = {
"__class__": "<class 'rl_agents.agents.tree_search.deterministic.DeterministicPlannerAgent'>",
"budget": 50,
"gamma": 0.7,
}
th_model = agent_factory(gym.make(args.env_name), agent_config)
time = False
elif args.env_name in ['duckietown']:
from a2c_ppo_acktr.duckietown.env import launch_env
from a2c_ppo_acktr.duckietown.wrappers import NormalizeWrapper, ImgWrapper,\
DtRewardWrapper, ActionWrapper, ResizeWrapper
from a2c_ppo_acktr.duckietown.teacher import PurePursuitExpert
env = launch_env()
env = ResizeWrapper(env)
env = NormalizeWrapper(env)
env = ImgWrapper(env)
env = ActionWrapper(env)
env = DtRewardWrapper(env)
# Create an imperfect demonstrator
expert = PurePursuitExpert(env=env)
time = False
else:
print('[Setting environemnt hyperparams variables]')
stats_path = os.path.join(args.rl_baseline_zoo_dir, 'trained_agents', f'{args.expert_algo}',\
f'{args.env_name}')
hyperparams, stats_path = get_saved_hyperparams(stats_path, test_mode=True,\
norm_reward=args.norm_reward_stable_baseline)
## Load saved policy
# subset of the environments have time wrapper
time_wrapper_envs = ['HalfCheetahBulletEnv-v0', 'Walker2DBulletEnv-v0', 'AntBulletEnv-v0']
if args.env_name in time_wrapper_envs:
time=True
else:
time = False
env = make_vec_envs(args.env_name, seed, 1, 0.99, f'{args.demo_data_dir}/tmp/gym', device,\
True, stats_path=stats_path, hyperparams=hyperparams, time=time)
th_model = Policy(
env.observation_space.shape,
env.action_space,
load_expert=True,
env_name=args.env_name,
rl_baseline_zoo_dir=args.rl_baseline_zoo_dir,
expert_algo=args.expert_algo,
# [Bug]: normalize=False,
normalize=True if hasattr(gym.envs, 'atari') else False,
base_kwargs={'recurrent': args.recurrent_policy}).to(device)
rtn_obs, rtn_acs, rtn_lens, ep_rewards = [], [], [], []
obs = env.reset()
if args.env_name in ['duckietown']:
obs = torch.FloatTensor([obs])
save = True
print(f'[running]')
step = 0
args.seed = args.seed
idx = random.randint(1,args.subsample_frequency)
obs_path_suffix = f'{args.demo_data_dir}/obs_{args.env_name}_seed={args.seed}'
acs_path_suffix = f'{args.demo_data_dir}/acs_{args.env_name}_seed={args.seed}'
while True:
with torch.no_grad():
if args.env_name in ['highway-v0']:
action = torch.tensor([[th_model.act(obs)]])
elif args.env_name in ['duckietown']:
action = torch.FloatTensor([expert.predict(None)])
elif hasattr(gym.envs, 'atari'):
_, actor_features, _ = th_model.base(obs, None, None)
# [Bug]: action = th.argmax(th_model.dist.linear(actor_features)).reshape(-1,1)
dist = th_model.dist(actor_features)
action = dist.sample()
else:
_, action, _, _ = th_model.act(obs, None, None, deterministic=True)
if isinstance(env.action_space, gym.spaces.Box):
clip_action = np.clip(action.cpu(), env.action_space.low, env.action_space.high)
else:
clip_action = action
if (step == idx and args.subsample) or not args.subsample:
#if args.env_name in env_hyperparam:
if time:
try: # If vectornormalize is on
rtn_obs.append(env.venv.get_original_obs())
except: # if vectornormalize is off
rtn_obs.append(env.venv.envs[0].get_original_obs())
else:
try: # If time is on and vectornormalize is on
rtn_obs.append(env.venv.get_original_obs())
except: # If time is off and vectornormalize is off
rtn_obs.append(obs.cpu().numpy().copy())
rtn_acs.append(action.cpu().numpy().copy())
idx += args.subsample_frequency
if args.env_name in ['duckietown']:
obs, reward, done, infos = env.step(clip_action.squeeze())
obs = torch.FloatTensor([obs])
else:
obs, reward, done, infos = env.step(clip_action)
step += 1
if args.env_name in ['duckietown']:
if done:
print(f"reward: {reward}")
ep_rewards.append(reward)
save = True
obs = env.reset()
obs = torch.FloatTensor([obs])
step = 0
idx=random.randint(1,args.subsample_frequency)
else:
for info in infos or done:
if 'episode' in info.keys():
print(f"reward: {info['episode']['r']}")
ep_rewards.append(info['episode']['r'])
save = True
obs = env.reset()
step = 0
idx=random.randint(1,args.subsample_frequency)
if (len(ep_rewards) in [1, 3, 5, 10, 15, 20]) and save:
rtn_obs_ = np.concatenate(rtn_obs)
rtn_acs_ = np.concatenate(rtn_acs)
obs_path = f'{obs_path_suffix}_ntraj={len(ep_rewards)}.npy'
acs_path = f'{acs_path_suffix}_ntraj={len(ep_rewards)}.npy'
print(f'saving to: {obs_path}')
print(f'saving to: {acs_path}')
np.save(obs_path, rtn_obs_)
np.save(acs_path, rtn_acs_)
print(f'done, length :{len(ep_rewards)}')
save = False
if len(ep_rewards) % 20 == 0:
break
print(f'expert: {np.mean(ep_rewards)}')
results_save_path = os.path.join(args.save_results_dir, 'expert', f'expert_{args.env_name}_seed={args.seed}.perf')
results = [{'total_num_steps':0 , 'train_loss': 0, 'test_loss': 0, 'num_trajs': 0 ,\
'test_reward':np.mean(ep_rewards), 'u_reward': 0}]
df = pd.DataFrame(results, columns=np.hstack(['x', 'steps', 'train_loss', 'test_loss',\
'train_reward', 'test_reward', 'label', 'u_reward']))
df.to_csv(results_save_path)