-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtransfer.py
135 lines (114 loc) · 4.93 KB
/
transfer.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
import numpy as np
from brain import brain
from GridWorld import GridWorld
class transfer(object):
def __init__(self):
pass
def from_to(self, agent : brain, state, state_):
agent.set_qvalue(state_, agent.get_q_table()[state])
return agent
if __name__ == '__main__':
env = GridWorld(9, 9, -1, 50, 100,150,1)
env.set_pick_up([2, 3, 4, 5, 6])
env.set_drop_off([18, 25, 27, 30, 34, 39, 43, 48, 110, 113, 119, 122, 133, 142, 145])
env.set_obstacles([19, 20, 22, 23, 26, 28, 29, 31, 32, 35, 37, 38, 40, 41, 44, \
46, 47, 49, 50, 53, 90, 91, 93, 94, 97, 98, 99, 100, 102, \
103, 106, 107, 108, 109, 111, 112, 115, 116, 117, 118, 120, \
121,124, 125, 149, 150, 151, 152, 153, 154, 155, 156, 158, 159,\
160, 161])
env.possible_states()
env.load_available_action2()
env.load_available_flag_dynamic2()
agent = brain(.1, .99, .1, len(env.action_space()), len(env.state_space()))
agent.load('qtable.txt')
# # #Primeiro Estagio
# # train_states = dict()
# # aux = []
# # for gp in env.get_possibles_grid_positions():
# # for pick in range(len(env.pick_up)):
# # for drop in range(len(env.drop_off)):
# # if gp not in {2, 3, 5, 6}:
# # aux.append(env.get_observation((0, 0, drop, pick, gp)))
# # train_states[env.get_observation((0, 0, 0, 2, gp))] = aux
# # aux = []
# # #Transferencia do conhecimento do primeiro estagio
# # transfer_learning = transfer()
# # for key in train_states.keys():
# # for state in train_states[key]:
# # agent = transfer_learning.from_to(agent, state = key, state_ = state)
# # agent.save('qtable2.txt')
# # # Primeiro Estagio
# # train_states = dict()
# # aux = []
# # for gp in env.get_possibles_grid_positions():
# # for pick in range(len(env.pick_up)):
# # for drop in range(len(env.drop_off)):
# # aux.append(env.get_observation((0, 0, drop, pick, gp)))
# # train_states[env.get_observation((0, 0, 0, pick, gp))] = aux
# # aux = []
# # # Transferencia do conhecimento do primeiro estagio
# # transfer_learning = transfer()
# # for key in train_states.keys():
# # for state in train_states[key]:
# # agent = transfer_learning.from_to(agent, state = key, state_ = state)
# # agent.save('qtable3.txt')
agent.load('qtable.txt')
# Chegar corretamento no pick para qualquer drop
# Primeiro Estagio
train_states = dict()
aux = []
for gp in env.get_possibles_grid_positions():
for pick in range(len(env.pick_up)):
for drop in range(len(env.drop_off)):
aux.append(env.get_observation((0, 2, drop, pick, gp)))
train_states[env.get_observation((0, 0, 0, 2, gp))] = aux
aux = []
# Transferencia do conhecimento do primeiro estagio
transfer_learning = transfer()
for key in train_states.keys():
for state in train_states[key]:
agent = transfer_learning.from_to(agent, state = key, state_ = state)
agent.save('qtable4.txt')
# # Primeiro Estagio - de Volta Pra Casa
# train_states = dict()
# aux = []
# for gp in env.get_possibles_grid_positions():
# for pick in range(len(env.pick_up)):
# for drop in range(len(env.drop_off)):
# if gp not in {2, 3, 5, 6}:
# aux.append(env.get_observation((0, 2, drop, pick, gp)))
# train_states[env.get_observation((0, 0, 0, 2, gp))] = aux
# aux = []
# # Transferencia do conhecimento do primeiro estagio
# transfer_learning = transfer()
# for key in train_states.keys():
# for state in train_states[key]:
# agent = transfer_learning.from_to(agent, state = key, state_ = state)
#
# agent.save('qtable.txt')
'''
train_states = dict()
aux = []
# Primeiro Estagio
for pick in range(len(env.pick_up)):
for gp in env.get_possibles_grid_positions():
for drop in range(len(env.drop_off)):
aux.append(env.get_observation((0, 0, drop, pick, gp)))
train_states[env.get_observation((0, 0, 0, pick, gp))] = aux[1:]
aux = []
# Transferencia do conhecimento do primeiro estagio
transfer_learning = transfer()
for key in train_states.keys():
for state in train_states[key]:
agent = transfer_learning.from_to(agent, state = key, state_ = state)
# Segundo Estagio
# Vale apenas (0, 1, drop, 0, [2, 3, 4, 5, 6])
train_states = dict()
aux = []
for drop in range(len(env.drop_off)):
for gp in env.pick_up:
for pick in range(len(env.pick_up)):
aux.append(env.get_observation((0, 1, drop, pick, gp)))
train_states[env.get_observation((0, 1, drop, 0, gp))] = aux[1:]
agent.save('qtable2.txt')
'''