-
Notifications
You must be signed in to change notification settings - Fork 1
/
csv_poses.py
194 lines (159 loc) · 8.06 KB
/
csv_poses.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
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
"""---------- Batchaya Noumeme Yacynte Divan ----------"""
"""---------- Bachelor's in Mechatronics Thesis ----------"""
"""---------- Development of Monocular Visual Odometry Algorithm, WS 23/24 ----------"""
"""---------- Technische Hochshule Wuerzburg-Schweinfurt ---------"""
"""---------- Centre for Robotics ---------"""
import csv
import numpy as np
import math
import cv2
from scipy.spatial.transform import Rotation
import matplotlib.pyplot as plt
from integrate import integrate
def imu_position(imu_data, t):
# Initialize velocity and position arrays
imu_data = np.array(imu_data)
velocity = imu_data[:,:3]
acceleration = imu_data[:,3:]
position = np.zeros(np.shape(acceleration))
'''for i in range(1,len(acceleration)):
velocity[i,:2] = acceleration[i,:2] * t[i] + velocity[i, :2]
velocity[i, 2] = (acceleration[i,2]+9.81) * t[i] + velocity[i, 2]'''
for i in range(1, len(t)):
position[i,0] = position[i-1,0] + velocity[i-1,0] * t[i]*2 - 0.5 * acceleration[i-1,0] * t[i]**2
position[i,1] = position[i-1,1] - velocity[i-1,1] * t[i] - 0.5 * acceleration[i-1,1] * t[i]**2
position[i,2] = position[i-1,2] + 0.5*(velocity[i-1,2] * t[i] + (acceleration[i-1,2] + 9.81) * t[i]**2)
"""# Plot results
plt.figure()
plt.plot(time_step, position[:,0])
plt.xlabel('Time (s)')
plt.ylabel('Position')
plt.title('Estimated Position from Accelerometer Data')
plt.grid(True)
plt.show()
print(nlfd)"""
return position
def get_opt_gt(filename, timestamps):
timestamp = []
with open(timestamps, newline='\n', encoding='utf-8') as csvfile:
file = csv.reader(csvfile, delimiter=',', quotechar='|')
k = 0
for time_stamp in file:
if k>0 :
timestamp.append((time_stamp[1],time_stamp[2],time_stamp[6], time_stamp[7], time_stamp[8], time_stamp[9],time_stamp[10], time_stamp[11], time_stamp[12] ))
k += 1
timestamp = np.array(timestamp, dtype=np.float64)
timestamp[:,-2] = timestamp[:,-2] + 981
with open(filename, newline='\n', encoding='utf-8') as csvfile:
rec = csv.reader(csvfile, delimiter=',', quotechar='|')
i = 0
j = 0
k = 0
poses = []
prev_row = np.zeros((6))
for row in rec:
if i>=7 and j < len(timestamp):
m = np.abs(np.float64(row[1]) - np.float64(timestamp[j,0]))
if (np.abs(np.float64(row[1]) - np.float64(timestamp[j,0])) < 0.01) and (not ('' in row[2:8]) ):
row = np.array(row[2:8])
row = np.float32(row)
x, y, z, t_x, t_y, t_z = row
# convert angles to rotation vector
optitrak__rotation = np.array([np.radians(x), np.radians(y), np.radians(z)])
# Assume you have translation vector in AirSim
optitrak_translation_vector = [t_x, t_y, t_z]
# Convert rotation vector to rotation matrix
# rotation_matrix_opencv = Rotation.from_euler('xyz', optitrak__rotation).as_matrix()
rotation_matrix_opencv, _ = cv2.Rodrigues(optitrak__rotation)
# The translation vector remains the same
translation_vector_opencv = np.array(optitrak_translation_vector)
# Stack R and t together to form the transformation matrix
R = np.array([[0,0,1],[0,1,0],[1,0,0]])
transformation_matrix_opencv = np.eye(4)
transformation_matrix_opencv[:3, :3] = rotation_matrix_opencv
transformation_matrix_opencv[:3, 3] = translation_vector_opencv
transformation_matrix_opencv[:3,:] = np.matmul(R,transformation_matrix_opencv[:3,:])
poses.append(transformation_matrix_opencv)
if len(poses) == 6:
print(optitrak__rotation, rotation_matrix_opencv)
j += 1
i +=1
poses_ = np.array(poses)
# print(poses_[5])
# Integrate acceleration from IMU to get the distances
dist = integrate((timestamp[:,2:8])/100, timestamp[:,0],poses_)
return poses, dist
def get_airsim_gt(filename):
def quaternion_to_euler(quaternion):
"""
Convert quaternion to Euler angles (roll, pitch, yaw).
:param quaternion: numpy array representing the quaternion [w, x, y, z]
:return: numpy array containing Euler angles [roll, pitch, yaw] in radians
"""
w, x, y, z = quaternion
# Roll (x-axis rotation)
sinr_cosp = 2.0 * (w * x + y * z)
cosr_cosp = 1.0 - 2.0 * (x * x + y * y)
roll = np.arctan2(sinr_cosp, cosr_cosp)
# Pitch (y-axis rotation)
sinp = 2.0 * (w * y - z * x)
if np.abs(sinp) >= 1:
pitch = np.sign(sinp) * np.pi / 2 # Use +/-90 degrees if out of range
else:
pitch = np.arcsin(sinp)
# Yaw (z-axis rotation)
siny_cosp = 2.0 * (w * z + x * y)
cosy_cosp = 1.0 - 2.0 * (y * y + z * z)
yaw = np.arctan2(siny_cosp, cosy_cosp)
return np.array([roll, pitch, yaw])
with open(filename,newline='\n',encoding='utf-8') as txtfile:
rec = csv.reader(txtfile, delimiter='\t', quotechar='|')
i = 0
poses = []
linear_acceleration = []
timestamp = []
for txtlist in rec:
if i>0:
pos = np.array(txtlist[2:9])
# acc = np.float32(np.array(txtlist[12:15])) + np.float32(np.array([0,0,9.81]))
times = float(txtlist[1])
timestamp.append(times)
pose_ = np.float32(pos)
t_x,t_y,t_z,w,x,y,z = pose_
# Assume you have rotation represented as a quaternion in AirSim
airsim_rotation_quaternion = [w, x, y, z]
# Assume you have translation vector in AirSim
airsim_translation_vector = [t_x, t_y, t_z]
# Convert quaternion to Euler angles
rotation_angles = quaternion_to_euler(airsim_rotation_quaternion)
# Convert Euler to rotation matrix
rotation_matrix, _ = cv2.Rodrigues(rotation_angles)
# rotation_matrix_opencv, _ = cv2.Rodrigues(np.array(airsim_rotation_quaternion[1:]))
# Convert quaternion to axis-angle representation and then to rotation matrix
# The translation vector remains the same
translation_vector_opencv = np.array(airsim_translation_vector)
# Stack R and t together to form the transformation matrix
R = np.array([[1,0,0],[0,0,1],[0,1,0]])
transformation_matrix_opencv = np.eye(4)
transformation_matrix_opencv[:3, :3] = rotation_matrix
transformation_matrix_opencv[:3, 3] = translation_vector_opencv
R1 = Rotation.from_euler('xyz', [0, 90, 0], degrees=True).as_matrix()
# transformation_matrix_opencv[:3,:] = np.matmul(R1,transformation_matrix_opencv[:3,:])
poses.append(transformation_matrix_opencv)
# acc = (R@acc.reshape(3,1)).reshape(3)
# linear_acceleration.append(acc)
i += 1
# print(poses)
R1 = Rotation.from_euler('xyz', [0, -90, 0], degrees=True).as_matrix()
linear_acceleration = np.array(linear_acceleration)
timestamp = np.array(timestamp)
timestamp = (timestamp - timestamp[0])/1000
poses_ = np.array(poses)
# position_of_imu = imu_position(linear_acceleration, timestamp)
poses_[:3,:3] = R1@poses_[:3,:3]
# dist = integrate_airsim(linear_acceleration, timestamp, poses_)
dist = []
for i in range(1,len(poses_)):
# d_vo.append(np.linalg.norm(vo[i]-vo[i-1]))
dist.append(np.linalg.norm(poses_[i,:3,3]-poses_[i-1,:3,3]))
return poses, dist