-
Notifications
You must be signed in to change notification settings - Fork 0
/
carpet_localiser.py
391 lines (318 loc) · 13.5 KB
/
carpet_localiser.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
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
#!/usr/bin/env python
from typing import Tuple, Optional
import pickle
import numpy as np
import cv2
import rospy
from std_msgs.msg import Header
from sensor_msgs.msg import Image
from geometry_msgs.msg import Quaternion, PoseArray, PoseWithCovarianceStamped
from geometry_msgs.msg import Pose as PoseMsg
from nav_msgs.msg import Odometry, OccupancyGrid
from cv_bridge import CvBridge
import message_filters
from tf.transformations import euler_from_quaternion, quaternion_from_euler
from carpet_color_classification import CarpetColorClassifier
from cbl_particle_filter.filter import CarpetBasedParticleFilter, Pose, OdomMeasurement, add_poses
import cbl_particle_filter.colors as colors
from cbl_particle_filter.carpet_map import load_map_from_png, CarpetMap
def yaw_from_quaternion_msg(q: Quaternion) -> float:
_, _, yaw = euler_from_quaternion([q.x, q.y, q.z, q.w])
return yaw
def quaternion_msg_from_yaw(yaw: float) -> Quaternion:
q_list = quaternion_from_euler(0, 0, yaw)
q = Quaternion(
x=q_list[0],
y=q_list[1],
z=q_list[2],
w=q_list[3],
)
return q
def odom_msg_to_measurement(odom_msg: Odometry) -> OdomMeasurement:
"""
Convert ROS odom to the cbl representation
"""
heading = yaw_from_quaternion_msg(odom_msg.pose.pose.orientation)
return OdomMeasurement(
odom_msg.pose.pose.position.x,
odom_msg.pose.pose.position.y,
heading,
)
def odom_msg_to_pose(odom_msg: Odometry) -> Pose:
"""
Convert ROS odom to the cbl pose representation
"""
heading = yaw_from_quaternion_msg(odom_msg.pose.pose.orientation)
return Pose(
odom_msg.pose.pose.position.x,
odom_msg.pose.pose.position.y,
heading,
)
def pose_to_pose_msg(pose: Pose) -> PoseMsg:
"""
convert cbl pose to ROS pose message
"""
pose_msg = PoseMsg()
pose_msg.position.x = pose.x
pose_msg.position.y = pose.y
pose_msg.orientation = quaternion_msg_from_yaw(pose.heading)
return pose_msg
def wrap_plus_minus_pi(angle):
return np.mod(angle + np.pi, 2 * np.pi) - np.pi
def compute_odom_delta(current_odom: OdomMeasurement,
previous_odom: OdomMeasurement) -> OdomMeasurement:
"""
return pose delta from prev to current, in prev frame
"""
if previous_odom is None:
return OdomMeasurement(0, 0, 0)
dx_global = current_odom.dx - previous_odom.dx
dy_global = current_odom.dy - previous_odom.dy
prev_heading = previous_odom.dheading
return OdomMeasurement(dx=dx_global * np.cos(prev_heading) +
dy_global * np.sin(prev_heading),
dy=-dx_global * np.sin(prev_heading) +
dy_global * np.cos(prev_heading),
dheading=wrap_plus_minus_pi(current_odom.dheading -
previous_odom.dheading))
def particles_to_pose_array(particles: np.ndarray) -> PoseArray:
"""
Convert particle filter particles into a ROS pose array message
"""
def particle_to_pose(particle: np.array) -> Pose:
return pose_to_pose_msg(Pose(*particle[0:3]))
return PoseArray(poses=[particle_to_pose(p) for p in particles])
def publish_image(map_png_file: str, cv_bridge: CvBridge,
pub: rospy.Publisher) -> None:
"""
Load the given image file and publish using the given publisher
"""
img = cv2.imread(map_png_file)
img_msg = cv_bridge.cv2_to_imgmsg(img, 'bgr8')
pub.publish(img_msg)
def write_color_name_on_cv_img(cv_img: np.ndarray,
color: colors.Color) -> np.ndarray:
"""
Write the name of the given color onto the image
"""
# define text colors
color_index_to_bgr_tuple = {
colors.BLACK.index: (0, 0, 0),
colors.LIGHT_BLUE.index: (255, 204, 51),
colors.BEIGE.index: (169, 214, 213),
colors.DARK_BLUE.index: (204, 51, 0),
colors.UNCLASSIFIED.index: (100, 100, 100)
}
cv2.rectangle(cv_img, (10, 220), (310, 150), (255, 255, 255),
-1) #background
return cv2.putText(img=cv_img,
text=color.name,
org=(10, 200),
fontFace=cv2.FONT_HERSHEY_SIMPLEX,
fontScale=1.6,
color=color_index_to_bgr_tuple[color.index],
thickness=10)
def publish_carpet_map_outline(carpet_map: CarpetMap, pub: rospy.Publisher):
"""
Publish an occupancy grid from the given carpet map.
The occupancy grid will show clear space where there is carpet, otherwise
occupied space
"""
o_grid = OccupancyGrid()
o_grid.header.frame_id = "map"
color_indices = [color.index for color in colors.COLORS]
o_grid.data = [
0 if elem in color_indices else 100
for elem in np.flipud(carpet_map.grid).flatten()
]
height, width = carpet_map.grid.shape
o_grid.info.height = height
o_grid.info.width = width
o_grid.info.resolution = carpet_map.cell_size
o_grid.info.origin.orientation.w = 1
pub.publish(o_grid)
class CarpetLocaliser():
"""
Interface between incoming ROS messages (odom and camera image) and
carpet based particle filter.
"""
def __init__(self,
map_png_file: str,
map_cell_size: float,
classifier_param_file: str,
log_inputs=False):
self.update_distance_threshold = 0.2 # m
self.update_rotation_threshold = 0.1 # rad
self.color_classifier = CarpetColorClassifier(classifier_param_file)
self.log_inputs = log_inputs
carpet = load_map_from_png(map_png_file, map_cell_size)
# create particle filter using params based on
# https://github.com/tim-fan/carpet_localisation/blob/master/notebooks/PF%20Parameter%20Optimisation%20-%205%20Office%20Loops.ipynb
# TODO: move params to yaml file
self.particle_filter = CarpetBasedParticleFilter(
carpet,
log_inputs,
resample_proportion=0,
weight_fn_p=0.9,
odom_pos_noise=0.01,
odom_heading_noise=0.02,
n_particles=500,
)
self.cv_bridge = CvBridge()
self.previous_odom = None
self.pose_pub = rospy.Publisher("current_pose",
Odometry,
queue_size=10)
self.particle_pub = rospy.Publisher(
"particlecloud",
PoseArray,
queue_size=10,
)
self.classified_image_pub = rospy.Publisher("classified_image",
Image,
queue_size=10)
# publish the carpet map on a latched image topic
self.map_pub = rospy.Publisher(
"carpet_map",
Image,
latch=True,
queue_size=10,
)
publish_image(map_png_file, self.cv_bridge, self.map_pub)
self.occupancy_pub = rospy.Publisher(
"carpet_map_outline",
OccupancyGrid,
latch=True,
queue_size=10,
)
publish_carpet_map_outline(carpet, self.occupancy_pub)
def __del__(self):
if self.log_inputs:
log_path = "/tmp/localiser_input_log.pickle"
self.particle_filter.write_input_log(log_path)
rospy.loginfo(f"Saved localiser input log to {log_path}")
def localisation_update(self,
odom_msg: Odometry,
img_msg: Image,
ground_truth_state: Optional[Odometry] = None):
"""
Update localisation based on given odom,image pair
Can optionally provide ground truth for logging purposes
"""
# prepare output pose message (odom)
current_pose = Odometry()
current_pose.header.frame_id = "map"
current_pose.header.stamp = img_msg.header.stamp
current_pose.twist = odom_msg.twist
# determine odom delta since last update
current_odom = odom_msg_to_measurement(odom_msg)
odom_delta = compute_odom_delta(current_odom, self.previous_odom)
# perform particle filter update, only if robot has travelled beyond
# a certain distance since the previous update
distance_since_last_update = np.sqrt(odom_delta.dx**2 +
odom_delta.dy**2)
rotation_since_last_update = np.abs(odom_delta.dheading)
if distance_since_last_update > self.update_distance_threshold or \
rotation_since_last_update > self.update_rotation_threshold or \
self.previous_odom is None:
rospy.loginfo(f"odom: {odom_delta}")
# determine detected color
color = self._classify_image_color(img_msg)
rospy.loginfo(f"detected color: {color.name}")
# get ground truth pose if provided
if ground_truth_state:
ground_truth_pose = odom_msg_to_pose(ground_truth_state)
rospy.loginfo(f"ground_truth: {ground_truth_pose}")
else:
ground_truth_pose = None
# perform update
self.particle_filter.update(odom_delta, color, ground_truth_pose)
# get current pose from the particle filter
current_pose.pose.pose = pose_to_pose_msg(
self.particle_filter.get_current_pose())
# also publish current particles
self._publish_particles(header=current_pose.header)
self.previous_odom = current_odom
else:
# set current pose as particle filter pose at last update
# plus accumulated odom since then
pf_pose = self.particle_filter.get_current_pose()
updated_pose_array = add_poses(
current_poses=np.array([[
pf_pose.x,
pf_pose.y,
pf_pose.heading,
]]),
pose_increments=np.array([[
odom_delta.dx,
odom_delta.dy,
odom_delta.dheading,
]])
)[0] # yapf: disable
updated_pose = Pose(x=updated_pose_array[0],
y=updated_pose_array[1],
heading=updated_pose_array[2])
current_pose.pose.pose = pose_to_pose_msg(updated_pose)
# publish current location
self.pose_pub.publish(current_pose)
def seed(self, pose_msg:PoseWithCovarianceStamped) -> None:
seed_pose = Pose(
x=pose_msg.pose.pose.position.x,
y=pose_msg.pose.pose.position.y,
heading=yaw_from_quaternion_msg(pose_msg.pose.pose.orientation),
)
self.particle_filter.seed(seed_pose)
self.previous_odom = None
self._publish_particles(header=pose_msg.header)
def _publish_particles(self, header:Header) -> None:
pose_array = particles_to_pose_array(
self.particle_filter.get_particles())
pose_array.header = header
self.particle_pub.publish(pose_array)
def _classify_image_color(self, img_msg: Image) -> colors.Color:
"""
invoke classifier on given image, returning color
"""
cv_image = self.cv_bridge.imgmsg_to_cv2(img_msg,
desired_encoding='bgr8')
_, color_name = self.color_classifier.classify(cv_image)
color = colors.color_from_name[color_name]
# republish the image with the classification written on it, for visualisation/debug
cv_image = write_color_name_on_cv_img(cv_image, color)
img_msg = self.cv_bridge.cv2_to_imgmsg(cv_image, encoding='bgr8')
self.classified_image_pub.publish(img_msg)
return color
def run_localisation():
rospy.init_node("carpet_localisation")
rospy.loginfo("initialised")
map_png_file = rospy.get_param("~map_png_file")
map_cell_size = rospy.get_param("~map_cell_size")
classifier_param_file = rospy.get_param("~classifier_param_file")
# param 'log_inputs': set true to record all inputs to the particle filter
# as a pickle file, for later offline playback
log_inputs = rospy.get_param("~log_inputs", default=False)
# param 'subscribe_ground_truth': set true to subscribe to ground truth pose
# (expected use-case = use with 'log_inputs' to log gazebo ground truth pose, for testing)
subscribe_ground_truth = rospy.get_param("~subscribe_ground_truth",
default=False)
localiser = CarpetLocaliser(map_png_file, map_cell_size,
classifier_param_file, log_inputs)
odom_sub = message_filters.Subscriber('odom', Odometry)
image_sub = message_filters.Subscriber('image', Image)
subs = [odom_sub, image_sub]
if subscribe_ground_truth:
ground_truth_sub = message_filters.Subscriber('ground_truth/state',
Odometry)
subs.append(ground_truth_sub)
time_synchronizer = message_filters.ApproximateTimeSynchronizer(
subs, 10, 0.2)
time_synchronizer.registerCallback(localiser.localisation_update)
init_pose_sub = rospy.Subscriber("initialpose", PoseWithCovarianceStamped, localiser.seed )
try:
rospy.spin()
except KeyboardInterrupt:
print("INTERRUPT!")
del localiser
print("DONE!")
if __name__ == '__main__':
run_localisation()