-
Notifications
You must be signed in to change notification settings - Fork 4
/
SLAM.py
153 lines (111 loc) · 4.8 KB
/
SLAM.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
import load_data as ld
import numpy as np
from SLAM_helper import *
import matplotlib.pyplot as plt
import MapUtils as maput
import cv2
import random
####Dataset####
joint = ld.get_joint("data/train_joint2")
lid = ld.get_lidar("data/train_lidar2")
itv = 1 # of skip in drawing maps
###############
angles = np.array([np.arange(-135, 135.25, 0.25) * np.pi / 180.0])
N, N_threshold = 100, 35
particles = np.zeros((N, 3))
weight = np.einsum('..., ...', 1.0 / N, np.ones((N, 1)))
mapfig = {}
mapfig['res'] = 0.05
mapfig['xmin'] = -40
mapfig['ymin'] = -40
mapfig['xmax'] = 40
mapfig['ymax'] = 40
mapfig['sizex'] = int(np.ceil((mapfig['xmax'] - mapfig['xmin']) / mapfig['res'] + 1))
mapfig['sizey'] = int(np.ceil((mapfig['ymax'] - mapfig['ymin']) / mapfig['res'] + 1))
mapfig['log_map'] = np.zeros((mapfig['sizex'], mapfig['sizey']))
mapfig['map'] = np.zeros((mapfig['sizex'], mapfig['sizey']), dtype = np.int8)
mapfig['show_map'] = 0.5 * np.ones((mapfig['sizex'], mapfig['sizey'], 3), dtype = np.int8)
pos_phy, posX_map, posY_map = {}, {}, {}
factor = np.array([1, 1, 10])
x_im = np.arange(mapfig['xmin'], mapfig['xmax'] + mapfig['res'], mapfig['res']) # x-positions of each pixel of the map
y_im = np.arange(mapfig['ymin'], mapfig['ymax'] + mapfig['res'], mapfig['res']) # y-positions of each pixel of the map
x_range = np.arange(-0.05, 0.06, 0.05)
y_range = np.arange(-0.05, 0.06, 0.05)
fig = plt.figure(1)
ax = fig.add_subplot(111)
ax.set_title("SLAM Map")
im = ax.imshow(mapfig['show_map'], cmap = "hot")
fig.show()
ts = joint['ts']
h_angle = joint['head_angles']
rpy_robot = joint['rpy']
lid_p = lid[0]
rpy_p = lid_p['rpy']
ind_0 = np.argmin(np.absolute(ts - lid_p['t'][0][0]))
pos_phy, posX_map, posY_map = mapConvert(lid_p['scan'], rpy_robot[:, ind_0], h_angle[:, ind_0], angles, particles, N, pos_phy, posX_map, posY_map, mapfig)
mapfig = drawMap(particles[0, :], posX_map[0], posY_map[0], mapfig)
pose_p, yaw_p = lid_p['pose'], rpy_p[0, 2]
timeline = len(lid)
for i in xrange(1, timeline):
print "{0}/{1}".format(i,timeline)
lid_c = lid[i]
pose_c, rpy_c = lid_c['pose'], lid_c['rpy']
yaw_c = rpy_c[0, 2]
yaw_est = particles[:, 2]
delta_x_gb = pose_c[0][0] - pose_p[0][0]
delta_y_gb = pose_c[0][1] - pose_p[0][1]
delta_theta_gb = yaw_c - yaw_p
delta_x_lc = np.einsum('..., ...', np.cos(yaw_p), delta_x_gb) + np.einsum('..., ...', np.sin(yaw_p), delta_y_gb)
delta_y_lc = np.einsum('..., ...', -np.sin(yaw_p), delta_x_gb) + np.einsum('..., ...', np.cos(yaw_p), delta_y_gb)
delta_theta_lc = delta_theta_gb
delta_x_gb_new = (np.einsum('..., ...', np.cos(yaw_est), delta_x_lc) - np.einsum('..., ...', np.sin(yaw_est), delta_y_lc)).reshape(-1, N)
delta_y_gb_new = (np.einsum('..., ...', np.sin(yaw_est), delta_x_lc) + np.einsum('..., ...', np.cos(yaw_est), delta_y_lc)).reshape(-1, N)
delta_theta_gb_new = np.tile(delta_theta_lc, (1, N))
ut = np.concatenate([np.concatenate([delta_x_gb_new, delta_y_gb_new], axis=0), delta_theta_gb_new], axis=0)
ut = np.einsum('ji', ut)
noise = np.einsum('..., ...', factor, np.random.normal(0, 1e-3, (N, 1)))
particles = particles + ut + noise
scan_c = lid_c['scan']
ind_i = np.argmin(np.absolute(ts - lid_c['t'][0][0]))
pos_phy, posX_map, posY_map = mapConvert(scan_c, rpy_robot[:, ind_i], h_angle[:, ind_i], angles, particles, N, pos_phy, posX_map, posY_map, mapfig)
corr = np.zeros((N, 1))
for i in range(N):
size = pos_phy[i].shape[1]
Y = np.concatenate([pos_phy[i], np.zeros((1, size))], axis = 0)
corr_cur = maput.mapCorrelation(mapfig['map'], x_im, y_im, Y[0 : 3, :], x_range, y_range)
ind = np.argmax(corr_cur)
corr[i] = corr_cur[ind / 3, ind % 3]
particles[i, 0] += x_range[ind / 3]
particles[i, 1] += y_range[ind % 3]
wtmp = np.log(weight) + corr
wtmp_max = wtmp[np.argmax(wtmp)]
lse = np.log(np.sum(np.exp(wtmp - wtmp_max)))
wtmp = wtmp - wtmp_max - lse
weight = np.exp(wtmp)
ind_best = weight.argmax()
x_r = (np.ceil((particles[ind_best, 0] - mapfig['xmin']) / mapfig['res']).astype(np.int16) - 1)
y_r = (np.ceil((particles[ind_best, 1] - mapfig['xmin']) / mapfig['res']).astype(np.int16) - 1)
mapfig['show_map'][x_r, y_r, 0] = 255
mapfig = drawMap(particles[ind_best, :], posX_map[ind_best], posY_map[ind_best], mapfig)
pose_p, yaw_p = pose_c, yaw_c
# Resampling
N_eff = 1 / np.sum(np.square(weight))
if N_eff < N_threshold:
#particles = resample(N, weight, particles)
particle_New = np.zeros((N, 3))
r = random.uniform(0, 1.0 / N)
c, i = weight[0], 0
for m in range(N):
u = r + m * (1.0 / N)
while u > c:
i = i + 1
c = c + weight[i]
particle_New[m, :] = particles[i, :]
particles = particle_New
weight = np.einsum('..., ...', 1.0 / N, np.ones((N, 1)))
#ax.imshow(mapfig['show_map'], cmap = "hot")
#im.set_data(mapfig['show_map'])
#im.axes.figure.canvas.draw()
fig1 = plt.figure(1)
plt.imshow(mapfig['show_map'], cmap = "hot")
plt.show()