-
Notifications
You must be signed in to change notification settings - Fork 3
/
Controller.py
167 lines (151 loc) · 8.55 KB
/
Controller.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
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
import numpy as np
import pandas as pd
from scipy.linalg import expm
from copy import deepcopy
class Controller(object):
def __init__(self, control_dim):
self.K = None
self.C = None
self.M = None
# initialize K,C,M parameters
self.control_dim = control_dim
# TODO 26 dim
action = np.loadtxt('/home/danieln7/Desktop/RobotCode2023/daniel_learning_runs/run3/action/scaled_params.csv', delimiter=',')
# action = np.loadtxt('daniel_params.csv', delimiter=',')
self.set_control_param(action=action)
def impedance_equation(self, pose_ref, vel_ref, pose_mod, vel_mod, f_int, f0, dt):
"""
Impedance Eq: F_int-F0=K(x0-xm)+C(x0_d-xm_d)-Mxm_dd
Solving the impedance equation for x(k+1)=Ax(k)+Bu(k) where
x(k+1)=[Xm,thm,Xm_d,thm_d]
Parameters:
x0,x0_d,th0,th0_d - desired goal position/orientation and velocity
F_int - measured force/moments in [N/Nm] (what the robot sense)
F0 - desired applied force/moments (what the robot does)
xm_pose - impedance model (updated in a loop) initialized at the initial pose of robot
A_d, B_d - A and B matrices of x(k+1)=Ax(k)+Bu(k)
Output:
X_nex = x(k+1) = [Xm,thm,Xm_d,thm_d]
"""
# state space formulation
# X=[xm;thm;xm_d;thm_d] U=[F_int;M_int;x0;th0;x0d;th0d]
A_1 = np.concatenate((np.zeros([6, 6], dtype=int), np.identity(6)), axis=1)
A_2 = np.concatenate((np.dot(-np.linalg.pinv(self.M), self.K), np.dot(-np.linalg.pinv(self.M), self.C)), axis=1)
A_temp = np.concatenate((A_1, A_2), axis=0)
B_1 = np.zeros([6, 18], dtype=int)
B_2 = np.concatenate((np.linalg.pinv(self.M), np.dot(np.linalg.pinv(self.M), self.K),
np.dot(np.linalg.pinv(self.M), self.C)), axis=1)
B_temp = np.concatenate((B_1, B_2), axis=0)
if np.isnan(A_temp).any() or np.isnan(B_temp).any():
s = 1
# discrete state space A, B matrices
A_d = expm(A_temp * dt)
B_d = np.dot(np.dot(np.linalg.pinv(A_temp), (A_d - np.identity(A_d.shape[0]))), B_temp)
# reference
x0 = pose_ref[:3].reshape(3, 1)
th0 = pose_ref[3:].reshape(3, 1)
x0_d = vel_ref[:3].reshape(3, 1)
th0_d = vel_ref[3:].reshape(3, 1)
# impedance model xm is initialized to initial position of the EEF and modified by force feedback
xm = pose_mod[:3].reshape(3, 1)
thm = pose_mod[3:].reshape(3, 1)
xm_d = vel_mod[:3].reshape(3, 1)
thm_d = vel_mod[3:].reshape(3, 1)
# State Space vectors
X = np.concatenate((xm, thm, xm_d, thm_d), axis=0) # 12x1 column vector
F = (f_int - f0).reshape(6, 1)
U = np.concatenate((F, x0, th0, x0_d, th0_d), axis=0).reshape(18, 1)
# discrete state solution X(k+1)=Ad*X(k)+Bd*U(k)
X_nex = np.dot(A_d, X) + np.dot(B_d, U)
# print(X_nex[9:12])
return X_nex.reshape(12, )
def set_control_param(self, action):
if self.control_dim == 26:
use_shir = False
if use_shir:
# # # Shirs Paper params:
# self.K = np.array([[9.90660954, 0., 0., 0., 46.63195038, 0.],
# [0., 33.75495148, 0., 157.51246643, 0., 0.],
# [0., 0., 18.89282036, 0., 0., 0.],
# [-38.88735199, 0., 0., 0., 32.1312713, 0.],
# [0., -17.89356422, 0., 79.34197998, 0., 0.],
# [0., 0., 0., 0., 0., 46.04693604]])
self.C = np.array([[6.16129827, 0., 0., 0., -1.56223333, 0.],
[0., 114.63842773, 0., 30.10368919, 0., 0.],
[0., 0., 2.75284362, 0., 0., 0.],
[0., -12.54157734, 0., 69.30596924, 0., 0.],
[-42.15202713, 0., 0., 0., 75.14640808, 0.],
[0., 0., 0., 0., 0., 26.27482986]])
self.M = np.array([[28.00367928, 0., 0., 0., 34.70161819, 0.],
[0., 71.05580902, 0., 37.04052734, 0., 0.],
[0., 0., 48.4661026, 0., 0., 0.],
[0., 39.43505096, 0., 63.75473022, 0., 0.],
[-44.1451416, 0., 0., 0., 7.56819868, 0.],
[0., 0., 0., 0., 0., 10.84090614]])
# Shirs Paper params with correction of K matrix:
self.K = np.array([[9.90660954, 0., 0., 0., 46.63195038, 0.],
[0., 33.75495148, 0., 157.51246643, 0., 0.],
[0., 0., 18.89282036, 0., 0., 0.],
[0., -17.89356422, 0., 79.34197998, 0., 0.],
[-38.88735199, 0., 0., 0., 32.1312713, 0.],
[0., 0., 0., 0., 0., 46.04693604]])
# #
else:
# print()
#
self.K = np.loadtxt('/home/danieln7/Desktop/RobotCode2023/daniel_learning_runs/run10/action/K.csv',
delimiter=',')
self.C = np.loadtxt('/home/danieln7/Desktop/RobotCode2023/daniel_learning_runs/run10/action/C.csv',
delimiter=',')
self.M = np.loadtxt('/home/danieln7/Desktop/RobotCode2023/daniel_learning_runs/run10/action/M.csv',
delimiter=',')
# print(self.K)
# print(self.C)
# print(self.M)
# self.K = np.array([[action[0], 0, 0, 0, action[1], 0],
# [0, action[2], 0, action[3], 0, 0],
# [0, 0, action[4], 0, 0, 0],
# [0, action[5], 0, action[6], 0, 0],
# [action[7], 0, 0, 0, action[8], 0],
# [0, 0, 0, 0, 0, action[9]]])
#
# self.C = np.array([[action[10], 0, 0, 0, action[11], 0],
# [0, action[12], 0, action[13], 0, 0],
# [0, 0, action[14], 0, 0, 0],
# [0, action[15], 0, action[16], 0, 0],
# [action[17], 0, 0, 0, action[18], 0],
# [0, 0, 0, 0, 0, action[19]]])
#
# self.M = np.array([[action[20], 0, 0, 0, 0, 0],
# [0, action[21], 0, 0, 0, 0],
# [0, 0, action[22], 0, 0, 0],
# [0, 0, 0, action[23], 0, 0],
# [0, 0, 0, 0, action[24], 0],
# [0, 0, 0, 0, 0, action[25]]])
# self.K = np.array([[abs(action[0]), 0, 0, 0, action[1], 0],
# [0, abs(action[2]), 0, action[3], 0, 0],
# [0, 0, abs(action[4]), 0, 0, 0],
# [0, action[5], 0, abs(action[6]), 0, 0],
# [action[7], 0, 0, 0, abs(action[8]), 0],
# [0, 0, 0, 0, 0, abs(action[9])]])
#
# self.C = np.array([[abs(action[10]), 0, 0, 0, action[11], 0],
# [0, abs(action[12]), 0, action[13], 0, 0],
# [0, 0, abs(action[14]), 0, 0, 0],
# [0, action[15], 0, abs(action[16]), 0, 0],
# [action[17], 0, 0, 0, abs(action[18]), 0],
# [0, 0, 0, 0, 0, abs(action[19])]])
#
# self.M = np.array([[abs(action[20]), 0, 0, 0, 0, 0],
# [0, abs(action[21]), 0, 0, 0, 0],
# [0, 0, abs(action[22]), 0, 0, 0],
# [0, 0, 0, abs(action[23]), 0, 0],
# [0, 0, 0, 0, abs(action[24]), 0],
# [0, 0, 0, 0, 0, abs(action[25])]])
print('---------------- K -----------------------')
print(self.K)
print('---------------- C -----------------------')
print(self.C)
print('---------------- M -----------------------')
print(self.M)
print()