-
Notifications
You must be signed in to change notification settings - Fork 0
/
UKFLengthVer.py
145 lines (112 loc) · 3.89 KB
/
UKFLengthVer.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
import numpy as np
import math
import scipy.io
import matplotlib.pyplot as plt
# mat_contents= scipy.io.loadmat('smallUnknown.mat') #l = .45
mat_contents= scipy.io.loadmat('smallKnown.mat') #pretty sure l = .38
accel = mat_contents['a_meas'][:,0]
gyro = mat_contents['d_meas'][:,0]
u = mat_contents['u'][:,0]
t = np.arange(0,20.001,0.01)
thetaTruth = mat_contents['th'][:,0]
n = 5
E = np.concatenate((math.sqrt(n)*np.eye(n),-math.sqrt(n)*np.eye(n)),axis = 1)
xHat = np.zeros((n,len(u)))
yBar = np.zeros((2,len(u)))
y = np.zeros((2,n*2))
P = .01 * np.eye(n)
# P[4,4] = .1
Q = np.zeros((n,n))
Q[0,0] = .0000001
Q[1,1] = .0000001
Q[2,2] = .000001
Q[3,3] = .000001
# Q[4,4] = .00000001
R = .0001 * np.eye(2)
m1 = .6
m2 = 1.9
# l = .38
# l = .4
la = .2
Ts = .01
xHat[4,0] = .4
xHat[0,0] = 3.14159
def f(state,u,steps = 10):
# print state
th = state[0]
# print th
thDot = state[1]
l = state[4]
# for i in range(steps):
# th = th + thDot*Ts/steps
# thDotDot = ((9.81*(m1+m2)-m1*l*math.cos(th)*thDot**2)*math.sin(th)+math.cos(th)*u)/(l*(m1+m2-m1*math.cos(th)**2))
# thDot = thDot + thDotDot*Ts/steps
# return np.array([th,thDot,state[2],state[3],state[4]])
thDotDot = ((9.81*(m1+m2)-m1*l*math.cos(th)*thDot**2)*math.sin(th)+math.cos(th)*u)/(l*(m1+m2-m1*math.cos(th)**2))
return np.array([th+thDot*Ts,thDot+thDotDot*Ts,state[2],state[3],state[4]])
def g(state,u):
th = state[0]
thDot = state[1]
l = state[4]
a = (l-la)/l*((math.cos(th)-m1*l*math.cos(th)*thDot**2*math.sin(th))*u/(m1+m2-m1*math.cos(th)**2)+(m1+m2)*9.81*math.sin(th)/(m1+m2-m1*math.cos(th)**2))
return np.array([thDot+state[2],a+state[3]])
for k in xrange(len(u)-1):
#measurement update
U, s, V = np.linalg.svd(P, full_matrices=True)
M = np.dot(U,np.diag(np.sqrt(s)))
xTilde = np.dot(M,E)
xTilde += xHat[:,k:k+1]
for i in xrange(2*n):
y[:,i] = g(xTilde[:,i],u[k])
yBar[:,k] = np.mean(y,1)
PXY = 1.0/(2.0*n)*np.dot(xTilde - xHat[:,k:k+1],np.transpose(y - yBar[:,k:k+1]))
PY = 1.0/(2.0*n)*np.dot(y - yBar[:,k:k+1],np.transpose(y-yBar[:,k:k+1]))+R
K = np.dot(PXY,np.linalg.inv(PY))
yMeas = np.reshape(np.array([gyro[k],accel[k]]),(2,1))
xHat[:,k:k+1] = xHat[:,k:k+1] + np.dot(K,yMeas-yBar[:,k:k+1])
P = P - np.dot(K,np.transpose(PXY))
#time update
U, s, V = np.linalg.svd(P, full_matrices=True)
M = np.dot(U,np.diag(np.sqrt(s)))
xTilde = np.dot(M,E)
xTilde += xHat[:,k:k+1]
for i in xrange(2*n):
xTilde[:,i] = f(xTilde[:,i],u[k])
xHat[:,k+1] = np.mean(xTilde,1)
P = 1.0/(2.0*n)*np.dot(xTilde - xHat[:,k+1:k+2],np.transpose(xTilde - xHat[:,k+1:k+2])) + Q
# f, axarr = plt.subplots(5, sharex=True)
# axarr[0].plot(t,u)
# axarr[0].plot(t,xHat[0,:])
# axarr[1].plot(t,xHat[1,:])
# axarr[2].plot(t,xHat[1,:]+xHat[2,:])
# axarr[2].plot(t,gyro)
# axarr[3].plot(t,xHat[2,:])
# axarr[3].plot(t,xHat[3,:])
# axarr[4].plot(t,xHat[4,:])
# plt.show()
f, axarr = plt.subplots(4, sharex=True)
# f.set_title('woo')
axarr[0].set_title('Small Input, Known Length')
# axarr[0].plot(t,u,label = 'Input [N*m]')
axarr[0].plot(t,xHat[0,:], label = 'Theta Est.[rad]')
axarr[0].plot(t,thetaTruth, label = 'Theta [rad]')
# axarr[0].set_title('In')
axarr[0].legend(fontsize = 10)
axarr[1].plot(t,xHat[1,:], label = 'Gyro Est. [rad/s]' )
axarr[1].plot(t,gyro, label = 'Gyro Meas. [rad/s]')
axarr[1].legend(fontsize = 10)
# axarr[2].plot(t,gyro)
axarr[2].plot(t,xHat[2,:],label = 'Gyro Bias [rad/s]')
axarr[2].plot(t,xHat[3,:], label = 'Accel. Bias [m/s^2]')
axarr[2].legend(fontsize = 10)
axarr[3].plot(t,xHat[4,:],label = 'Length Estimate')
axarr[3].legend(fontsize = 10,loc = 4)
plt.xlabel('time [s]')
axarr[0].grid(b=True, which='major', color='black', linestyle='-')
axarr[1].grid(b=True, which='major', color='black', linestyle='-')
axarr[2].grid(b=True, which='major', color='black', linestyle='-')
axarr[3].grid(b=True, which='major', color='black', linestyle='-')
# plt.show()
plt.subplots_adjust(left=0.08, right=.95, top=0.95, bottom=0.05)
# f.figsize=(20,10)
plt.show()