-
Notifications
You must be signed in to change notification settings - Fork 1
/
run_sensitivity.py
110 lines (91 loc) · 3.15 KB
/
run_sensitivity.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
from fuzzy_set import FuzzySet
from ARIC_model import ARICModel
import quantizations
from utils import sgn, plot_fuzzy_memberships
import matplotlib.pyplot as plt
import numpy as np
from pole_cart import PoleCart
# Rule set: logic statements of the form IF [C1] AND [C2] THEN [C3]
# The conditions (C1 and C2) are composed of the index of the input
# variable and a fuzzy set. The conclusions (C3) are simply fuzzy sets.
rule_set = [
[(2, "NE"), (3, "NE"), "NL"],
[(2, "NE"), (3, "ZE"), "NS"],
[(2, "NE"), (3, "PO"), "0"],
[(2, "ZE"), (3, "NE"), "NM"],
[(2, "ZE"), (3, "ZE"), "0"],
[(2, "ZE"), (3, "PO"), "PM"],
[(2, "PO"), (3, "NE"), "0"],
[(2, "PO"), (3, "ZE"), "PS"],
[(2, "PO"), (3, "PO"), "PL"],
[(0, "NE"), (1, "NE"), "NS"],
[(0, "VS"), (1, "NE"), "NVS"],
[(0, "VS"), (1, "PO"), "PVS"],
[(0, "PO"), (1, "PO"), "PS"]
]
def o_func_cart_pole(v, p):
import random
q = (p + 1) / 2
return v if random.random() < q else -v
def k_func_cart_pole(v, v_prev, p):
return 1 - p if sgn(v) != sgn(v_prev) else -p
# <START DEBUG CODE>
# aric.show_imf()
# aric.show_omf()
# env = PoleCart([0, 0.0, 0.0, 0], dt=0.001)
# env.sim_and_plot(10)
# exit()
# <END DEBUG CODE>
max_trials = 1000
show_pole_cart = False
show_plots = True
use_gym = False
init_state = [0, 0.0, 0.01, 0] # [x, xd, theta, thetad]
dt = 0.02
env = PoleCart(init_state, dt=dt)
# STEP 1: TRAIN CONTROLLER IN NORMAL CONDITIONS
aric = ARICModel(4, quantizations.berenji_quantization_inputs, quantizations.berenji_quantization_outputs, rule_set,
o_func_cart_pole, k_func_cart_pole, discount_rate=0.9, beta=0.2, beta_h=0.05, rho=1.0, rho_h=0.2)
trial_durations = []
for i in range(200):
done = False
env.reset()
its = 0
while not done:
y = env.step() # comes out as [x, x_dot, theta, theta_dot]
done = np.rad2deg(abs(y[2])) > 12 or abs(y[0]) > 2.4
control_input = 0.1 * aric.process_state_input(y, status="operating" if not done else "fail")
env.f = control_input
its += 1
if its > 500000:
done = True
print(f"Trial #{i}/{200} done in {its} steps")
if its > 500000:
break
# STEP 2: TEST CONTROLLER IN DIFFERENT CONDITIONS
trial_durations = []
scale_factors = [0.25, 0.5, 0.75, 1.25, 1.75, 2.5, 3.5, 4.5, 5.5]
for scale_factor in scale_factors:
print(f"Sensitivity test for factor {scale_factor}")
env.l = 0.5 * scale_factor
env.m = 0.1 * scale_factor
env.m_c = 1.0 * scale_factor
env.reset()
its = 0
done = False
while not done:
y = env.step() # comes out as [x, x_dot, theta, theta_dot]
control_input = 0.1 * aric.process_state_input(y, status="operating" if not done else "fail")
env.f = control_input
its += 1
done = np.rad2deg(abs(y[2])) > 12 or abs(y[0]) > 2.4 or its > 200000
print(f'\tFailed in {its} steps')
trial_durations.append(its)
plt.plot(scale_factors, trial_durations, color='k')
plt.xlabel('Scale factor')
plt.plot([0, 200], [100000, 100000], 'g--')
plt.ylabel('Time steps (20 ms)')
plt.ticklabel_format(style='sci', axis='y', scilimits=(0, 3))
plt.grid()
# plt.legend()
plt.show()