forked from djansr1408/AtariGamePong
-
Notifications
You must be signed in to change notification settings - Fork 0
/
utils.py
47 lines (43 loc) · 1.63 KB
/
utils.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
import numpy as np
def preprocess_frame(I):
"""
Preprocess input image(frame) of shape 210x160x3 into 6400(80x80) 1D float vector.
:param I: input frame of shape 210x160x3 uint8
:return: 1D flaot vector of length 6400
"""
I = I[35:195]
I = I[::2, ::2, 0] # downsample by factor of 2
I[I == 144] = 0 # erase background (type 1)
I[I == 109] = 0 # erase background (type 2)
I[I != 0] = 1 # everything else (paddles, ball) just set to 1
return I.astype(np.float).ravel()
# def discount_rewards(rewards, gamma):
# """
# Takes 1D array/list of rewards and computes discounted reward.
# :param r: 1D array of rewards
# :param gamma: Discount factor
# :return: Discounted reward array/list
# """
# cumulative_reward = 0
# dis_rewards = np.zeros(len(rewards), dtype=np.float32)
# for t in reversed(range(len(rewards))):
# if rewards[t] != 0: # this is just for Pong game!!!
# cumulative_reward = 0
# cumulative_reward = gamma * cumulative_reward + rewards[t]
# dis_rewards[t] = cumulative_reward
# return dis_rewards
def discount_rewards(r, gamma):
"""
Takes 1D array of rewards and computes discounted reward.
:param r: 1D array of rewards
:return: discounted reward
"""
r = np.array(r)
discount_r = np.zeros_like(r, dtype=np.float32)
running_add = 0
for t in reversed(range(0, len(r))):
if r[t] != 0:
running_add = 0 # reset the sum, since this was a game boundary(Pong specific!)
running_add = running_add * gamma + r[t]
discount_r[t] = running_add
return discount_r