File size: 1,790 Bytes
ebb75df
6aec4d8
 
 
ebb75df
a2ff203
ebb75df
6aec4d8
a2ff203
ebb75df
 
a2ff203
6aec4d8
ebb75df
6aec4d8
 
 
 
 
36f3504
 
 
6aec4d8
ebb75df
 
 
6aec4d8
 
 
 
 
36f3504
 
6aec4d8
 
36f3504
6aec4d8
a2ff203
 
 
 
 
 
 
 
 
6aec4d8
a2ff203
 
 
 
6aec4d8
ebb75df
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
import gymnasium as gym
from stable_baselines3 import DQN
from stable_baselines3.common.monitor import Monitor
from stable_baselines3.common.vec_env import VecVideoRecorder, DummyVecEnv, VecEnv

model_name = "dqn_v2-5/ALE-Pacman-v5" # path to model, should be an argument
env_id = "ALE/Pacman-v5"
video_folder = "videos/"
video_length = 100000 #steps

vec_env = DummyVecEnv([lambda: gym.make(env_id, render_mode="rgb_array")])
model = DQN.load(model_name)
# output: <stable_baselines3.common.vec_env.dummy_vec_env.DummyVecEnv object at 0x0000029974DC6550>

# vec_env = gym.make(env_id, render_mode="rgb_array")
# output <OrderEnforcing<PassiveEnvChecker<AtariEnv<ALE/Pacman-v5>>>>

# vec_env = Monitor(gym.make(env_id, render_mode="rgb_array"))

print("\n\n\n")
print(vec_env)
print("\n\n\n")

obs = vec_env.reset()


# Record the video starting at the first step
vec_env = VecVideoRecorder(vec_env,
                           video_folder, 
                           record_video_trigger=lambda x: x == 0,
                           video_length=video_length,
                          #  name_prefix=f"video-{env_id}"
                           )
#  Once I make the environment, now I need to walk through it...???
#   I want to act according to the policy that has been trained
obs = vec_env.reset()
print(vec_env)
# for _ in range(video_length + 1):
#   action, states = model.predict(obs)
#   obs, _, _, _ = vec_env.step(action)

# Instead of using the specified steps in a for loop
# use a while loop to check if the episode has terminated
# Stop recording when the episode ends
end = True
while end == True:
  action, states = model.predict(obs)
  obs, _, done, _ = vec_env.step(action)
  if done == True:
    print("exiting loop")
    end = False
# # Save the video
vec_env.close()