-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathactor_network.py
147 lines (116 loc) · 6.06 KB
/
actor_network.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
import tensorflow as tf
import numpy as np
import math
# Hyper Parameters
LAYER1_SIZE = 300
LAYER2_SIZE = 400
LEARNING_RATE = 1e-4
TAU = 0.001
BATCH_SIZE = 32
class ActorNetwork:
"""docstring for ActorNetwork"""
def __init__(self, sess, state_dim, action_dim, alextmodel_actor):
self.sess = sess
self.state_dim = state_dim
self.action_dim = action_dim
# create actor network
self.state_input, self.image_input, self.action_output, self.net = self.create_network(state_dim, action_dim,
alextmodel_actor)
# create target actor network
self.target_state_input, self.target_image_input,self.target_action_output, self.target_update, self.target_net = self.create_target_network(
state_dim, action_dim, self.net, alextmodel_actor)
# define training rules
self.create_training_method()
self.sess.run(tf.initialize_all_variables())
self.update_target()
# self.load_network()
def create_training_method(self):
self.q_gradient_input = tf.placeholder("float", [None, self.action_dim])
self.parameters_gradients = tf.gradients(self.action_output, self.net, -self.q_gradient_input)
'''
for i, grad in enumerate(self.parameters_gradients):
if grad is not None:
self.parameters_gradients[i] = tf.clip_by_value(grad, -2.0,2.0)
'''
self.optimizer = tf.train.AdamOptimizer(LEARNING_RATE).apply_gradients(zip(self.parameters_gradients, self.net))
def create_network(self, state_dim, action_dim, alextmodel_actor):
layer1_size = LAYER1_SIZE
layer2_size = LAYER2_SIZE
state_input = tf.placeholder("float", [None, state_dim])
image_input = tf.placeholder(tf.float32, shape=[None, 64, 64, 3])
cnn_output = alextmodel_actor(image_input)
W1 = self.variable([state_dim, layer1_size], state_dim)
b1 = self.variable([layer1_size], state_dim)
W2 = self.variable([layer1_size, layer2_size], layer1_size)
b2 = self.variable([layer2_size], layer1_size)
# W3 = tf.Variable(tf.random_uniform([layer2_size,action_dim],-3e-3,3e-3))
# b3 = tf.Variable(tf.random_uniform([action_dim],-3e-3,3e-3))
W_steer = tf.Variable(tf.random_uniform([layer2_size + 512, 1], -1e-4, 1e-4))
b_steer = tf.Variable(tf.random_uniform([1], -1e-4, 1e-4))
W_accel = tf.Variable(tf.random_uniform([layer2_size + 512, 1], -1e-4, 1e-4))
b_accel = tf.Variable(tf.random_uniform([1], -1e-4, 1e-4))
W_brake = tf.Variable(tf.random_uniform([layer2_size + 512, 1], -1e-4, 1e-4))
b_brake = tf.Variable(tf.random_uniform([1], -1e-4, 1e-4))
layer1 = tf.nn.relu(tf.matmul(state_input, W1) + b1)
layer2 = tf.nn.relu(tf.matmul(layer1, W2) + b2)
layer2 = tf.concat([layer2, cnn_output], 1)
steer = tf.tanh(tf.matmul(layer2, W_steer) + b_steer)
accel = tf.sigmoid(tf.matmul(layer2, W_accel) + b_accel)
brake = tf.sigmoid(tf.matmul(layer2, W_brake) + b_brake)
# action_output = tf.concat(1, [steer, accel, brake])
action_output = tf.concat([steer, accel, brake], 1)
return state_input, image_input, action_output, [W1, b1, W2, b2, W_steer, b_steer, W_accel, b_accel, W_brake,
b_brake]
def create_target_network(self, state_dim, action_dim, net, alextmodel_actor):
state_input = tf.placeholder("float", [None, state_dim])
image_input = tf.placeholder(tf.float32, shape=[None, 64, 64, 3])
cnn_output = alextmodel_actor(image_input)
a_params = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, scope='actor')
net.extend(a_params)
ema = tf.train.ExponentialMovingAverage(decay=1 - TAU)
target_update = ema.apply(net)
target_net = [ema.average(x) for x in net]
layer1 = tf.nn.relu(tf.matmul(state_input, target_net[0]) + target_net[1])
layer2 = tf.nn.relu(tf.matmul(layer1, target_net[2]) + target_net[3])
layer2 = tf.concat([layer2, cnn_output], 1)
steer = tf.tanh(tf.matmul(layer2, target_net[4]) + target_net[5])
accel = tf.sigmoid(tf.matmul(layer2, target_net[6]) + target_net[7])
brake = tf.sigmoid(tf.matmul(layer2, target_net[8]) + target_net[9])
# action_output = tf.concat(1, [steer, accel, brake])
action_output = tf.concat([steer, accel, brake], 1)
return state_input, image_input, action_output, target_update, target_net
def update_target(self):
self.sess.run(self.target_update)
def train(self, q_gradient_batch, state_batch):
self.sess.run(self.optimizer, feed_dict={
self.q_gradient_input: q_gradient_batch,
self.state_input: state_batch
})
def actions(self, state_batch):
return self.sess.run(self.action_output, feed_dict={
self.state_input: state_batch
})
def action(self, state):
return self.sess.run(self.action_output, feed_dict={
self.state_input: [state]
})[0]
def target_actions(self, state_batch):
return self.sess.run(self.target_action_output, feed_dict={
self.target_state_input: state_batch
})
# f fan-in size
def variable(self, shape, f):
return tf.Variable(tf.random_uniform(shape, -1 / math.sqrt(f), 1 / math.sqrt(f)))
'''
def load_network(self):
self.saver = tf.train.Saver()
checkpoint = tf.train.get_checkpoint_state("saved_actor_networks")
if checkpoint and checkpoint.model_checkpoint_path:
self.saver.restore(self.sess, checkpoint.model_checkpoint_path)
print "Successfully loaded:", checkpoint.model_checkpoint_path
else:
print "Could not find old network weights"
def save_network(self,time_step):
print 'save actor-network...',time_step
self.saver.save(self.sess, 'saved_actor_networks/' + 'actor-network', global_step = time_step)
'''