-
Notifications
You must be signed in to change notification settings - Fork 34
/
register_camera.py
398 lines (344 loc) · 18.4 KB
/
register_camera.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
392
393
394
395
396
397
398
"""
Script to register sensors to a chessboard for the YuMi setup
Authors: Jeff Mahler and Brenton Chu
"""
import argparse
import cv2
import IPython
import logging
import numpy as np
import os
import sys
import time
import traceback
from mpl_toolkits.mplot3d import Axes3D
import rospy
import rosgraph.roslogging as rl
from autolab_core import Point, PointCloud, RigidTransform, YamlConfig
from perception import CameraChessboardRegistration, RgbdSensorFactory
from visualization import Visualizer2D as vis2d
from visualization import Visualizer3D as vis3d
from yumipy import YuMiRobot
from yumipy import YuMiConstants as YMC
global clicked_pt
clicked_pt = None
pt_radius = 2
pt_color = (255,0,0)
def click_gripper(event, x, y, flags, param):
global clicked_pt
if event == cv2.EVENT_LBUTTONDBLCLK:
clicked_pt = np.array([x,y])
logging.info('Clicked: {}'.format(clicked_pt))
if __name__ == '__main__':
logging.getLogger().setLevel(logging.INFO)
# parse args
parser = argparse.ArgumentParser(description='Register a camera to a robot')
parser.add_argument('--config_filename', type=str, default='cfg/tools/register_camera.yaml', help='filename of a YAML configuration for registration')
args = parser.parse_args()
config_filename = args.config_filename
config = YamlConfig(config_filename)
# get known tf from chessboard to world
T_cb_world = RigidTransform.load(config['chessboard_tf'])
# initialize node
rospy.init_node('register_camera', anonymous=True)
logging.getLogger().addHandler(rl.RosStreamHandler())
# get camera sensor object
for sensor_frame, sensor_data in config['sensors'].iteritems():
logging.info('Registering %s' %(sensor_frame))
sensor_config = sensor_data['sensor_config']
registration_config = sensor_data['registration_config'].copy()
registration_config.update(config['chessboard_registration'])
# open sensor
try:
sensor_type = sensor_config['type']
sensor_config['frame'] = sensor_frame
logging.info('Creating sensor')
sensor = RgbdSensorFactory.sensor(sensor_type, sensor_config)
logging.info('Starting sensor')
sensor.start()
ir_intrinsics = sensor.ir_intrinsics
logging.info('Sensor initialized')
# register
reg_result = CameraChessboardRegistration.register(sensor, registration_config)
T_camera_world = T_cb_world * reg_result.T_camera_cb
logging.info('Final Result for sensor %s' %(sensor_frame))
logging.info('Rotation: ')
logging.info(T_camera_world.rotation)
logging.info('Quaternion: ')
logging.info(T_camera_world.quaternion)
logging.info('Translation: ')
logging.info(T_camera_world.translation)
except Exception as e:
logging.error('Failed to register sensor {}'.format(sensor_frame))
traceback.print_exc()
continue
# fix the chessboard corners
if config['fix_orientation_cb_to_world']:
# read params
num_pts_x = config['grid_x']
num_pts_y = config['grid_y']
grid_width = config['grid_width']
grid_height = config['grid_height']
gripper_height = config['gripper_height']
grid_center_x = config['grid_center_x']
grid_center_y = config['grid_center_y']
# determine robot poses
robot_poses = []
for i in range(num_pts_x):
x = -float(grid_width) / 2 + grid_center_x + float(i * grid_width) / num_pts_x
for j in range(num_pts_y):
y = -float(grid_height) / 2 + grid_center_y + float(j * grid_height) / num_pts_y
# form robot pose
R_robot_world = np.array([[1, 0, 0],
[0, 0, 1],
[0, -1, 0]])
t_robot_world = np.array([x, y, gripper_height])
T_robot_world = RigidTransform(rotation=R_robot_world,
translation=t_robot_world,
from_frame='gripper',
to_frame='world')
robot_poses.append(T_robot_world)
# start robot
y = YuMiRobot(tcp=YMC.TCP_SUCTION_STIFF)
y.set_z('fine')
y.reset_home()
global clicked_pt
# iteratively go to poses
robot_points_camera = []
for robot_pose in robot_poses:
# reset clicked pt
clicked_pt = None
# move to pose
y.right.goto_pose(robot_pose, wait_for_res=True)
# capture image
color_im, depth_im, _ = sensor.frames()
depth_im = depth_im.inpaint(0.25)
cv2.namedWindow('click')
cv2.setMouseCallback('click', click_gripper)
while True:
if clicked_pt is None:
cv2.imshow('click', color_im.data)
else:
im = color_im.data.copy()
cv2.circle(im,
tuple(clicked_pt.tolist()),
pt_radius,
pt_color,
-1)
cv2.imshow('click', im)
key = cv2.waitKey(1) & 0xFF
if key == ord('q') and clicked_pt is not None:
logging.info('Moving to next pose')
break
# store clicked pt in 3D
logging.info('Point collection complete')
depth = depth_im[clicked_pt[1], clicked_pt[0]]
p = ir_intrinsics.deproject_pixel(depth,
Point(clicked_pt, frame=ir_intrinsics.frame))
robot_points_camera.append(p.data)
# reset
y.reset_home()
y.stop()
# collect
true_robot_points_world = PointCloud(np.array([T.translation for T in robot_poses]).T,
frame=ir_intrinsics.frame)
est_robot_points_world = T_camera_world * PointCloud(np.array(robot_points_camera).T,
frame=ir_intrinsics.frame)
mean_true_robot_point = np.mean(true_robot_points_world.data, axis=1).reshape(3,1)
mean_est_robot_point = np.mean(est_robot_points_world.data, axis=1).reshape(3,1)
# fit a plane
best_R_cb_world = None
best_dist = np.inf
k = 0
K = 25
num_poses = len(robot_poses)
sample_size = int(num_poses * 0.3)
min_inliers = int(num_poses * 0.6)
dist_thresh = 0.0015
true_robot_points_world._data = true_robot_points_world._data - mean_true_robot_point
est_robot_points_world._data = est_robot_points_world._data - mean_est_robot_point
while k < K:
ind = np.random.choice(num_poses, size=sample_size, replace=False)
H = est_robot_points_world.data[:,ind].dot(true_robot_points_world.data[:,ind].T)
U, S, V = np.linalg.svd(H)
R_cb_world = V.T.dot(U.T)
fixed_robot_points_world = R_cb_world.dot(est_robot_points_world.data)
diffs = fixed_robot_points_world - true_robot_points_world.data
dists = np.linalg.norm(diffs, axis=0)
inliers = np.where(dists < dist_thresh)[0]
num_inliers = inliers.shape[0]
print k, num_inliers, np.mean(dists)
if num_inliers >= min_inliers:
ind = inliers
H = est_robot_points_world.data[:,ind].dot(true_robot_points_world.data[:,ind].T)
U, S, V = np.linalg.svd(H)
R_cb_world = V.T.dot(U.T)
fixed_robot_points_world = R_cb_world.dot(est_robot_points_world.data)
diffs = fixed_robot_points_world - true_robot_points_world.data
dists = np.linalg.norm(diffs, axis=0)
mean_dist = np.mean(dists[ind])
if mean_dist < best_dist:
best_dist = mean_dist
best_R_cb_world = R_cb_world
k += 1
R_cb_world = best_R_cb_world
T_corrected_cb_world = RigidTransform(rotation=R_cb_world,
from_frame='world',
to_frame='world')
R_cb_world = R_cb_world.dot(T_cb_world.rotation)
T_cb_world = RigidTransform(rotation=R_cb_world,
translation=T_cb_world.translation,
from_frame=T_cb_world.from_frame,
to_frame=T_cb_world.to_frame)
T_camera_world = T_cb_world * reg_result.T_camera_cb
T_cb_world.save(config['chessboard_tf'])
# vis
if config['vis_points']:
_, depth_im, _ = sensor.frames()
points_world = T_camera_world * ir_intrinsics.deproject(depth_im)
true_robot_points_world = PointCloud(np.array([T.translation for T in robot_poses]).T,
frame=ir_intrinsics.frame)
est_robot_points_world = T_camera_world * PointCloud(np.array(robot_points_camera).T,
frame=ir_intrinsics.frame)
mean_est_robot_point = np.mean(est_robot_points_world.data, axis=1).reshape(3,1)
est_robot_points_world._data = est_robot_points_world._data - mean_est_robot_point + mean_true_robot_point
fixed_robot_points_world = T_corrected_cb_world * est_robot_points_world
mean_fixed_robot_point = np.mean(fixed_robot_points_world.data, axis=1).reshape(3,1)
fixed_robot_points_world._data = fixed_robot_points_world._data - mean_fixed_robot_point + mean_true_robot_point
vis3d.figure()
vis3d.points(points_world, color=(0,1,0), subsample=10, random=True, scale=0.001)
vis3d.points(true_robot_points_world, color=(0,0,1), scale=0.001)
vis3d.points(fixed_robot_points_world, color=(1,1,0), scale=0.001)
vis3d.points(est_robot_points_world, color=(1,0,0), scale=0.001)
vis3d.pose(T_camera_world)
vis3d.show()
# save tranformation arrays based on setup
output_dir = os.path.join(config['calib_dir'], sensor_frame)
if not os.path.exists(output_dir):
os.makedirs(output_dir)
pose_filename = os.path.join(output_dir, '%s_to_world.tf' %(sensor_frame))
T_camera_world.save(pose_filename)
intr_filename = os.path.join(output_dir, '%s.intr' %(sensor_frame))
ir_intrinsics.save(intr_filename)
f = open(os.path.join(output_dir, 'corners_cb_%s.npy' %(sensor_frame)), 'w')
np.save(f, reg_result.cb_points_cam.data)
# move the robot to the chessboard center for verification
if config['use_robot']:
# get robot type
robot_type = 'yumi'
if 'robot_type' in config.keys():
robot_type = config['robot_type']
# find the rightmost and further cb point in world frame
cb_points_world = T_camera_world * reg_result.cb_points_cam
cb_point_data_world = cb_points_world.data
dir_world = np.array([-1.0, 1.0, 0])
dir_world = dir_world / np.linalg.norm(dir_world)
ip = dir_world.dot(cb_point_data_world)
# open interface to robot
if robot_type == 'ur5':
from ur_control import UniversalRobot, ToolState, T_KINEMATIC_AVOIDANCE_WORLD, KINEMATIC_AVOIDANCE_JOINTS
robot = UniversalRobot()
robot.reset_home()
else:
y = YuMiRobot(tcp=YMC.TCP_SUCTION_STIFF)
y.reset_home()
robot = y.right
waypoints = []
time.sleep(1)
# choose target point #1
target_ind = np.where(ip == np.max(ip))[0]
target_pt_world = cb_points_world[target_ind[0]]
# create robot pose relative to target point
if robot_type == 'ur5':
R_gripper_world = np.array([[-1.0, 0, 0],
[0, 1.0, 0],
[0, 0, -1.0]])
else:
R_gripper_world = np.array([[1.0, 0, 0],
[0, -1.0, 0],
[0, 0, -1.0]])
t_gripper_world = np.array([target_pt_world.x + config['gripper_offset_x'],
target_pt_world.y + config['gripper_offset_y'],
target_pt_world.z + config['gripper_offset_z']])
T_gripper_world = RigidTransform(rotation=R_gripper_world,
translation=t_gripper_world,
from_frame='gripper',
to_frame='cb')
logging.info('Moving robot to point x=%f, y=%f, z=%f' %(t_gripper_world[0], t_gripper_world[1], t_gripper_world[2]))
T_lift = RigidTransform(translation=(0,0,0.05), from_frame='cb', to_frame='cb')
T_gripper_world_lift = T_lift * T_gripper_world
T_orig_gripper_world_lift = T_gripper_world_lift.copy()
if config['vis_cb_corners']:
_, depth_im, _ = sensor.frames()
points_world = T_camera_world * ir_intrinsics.deproject(depth_im)
vis3d.figure()
vis3d.points(cb_points_world, color=(0,0,1), scale=0.005)
vis3d.points(points_world, color=(0,1,0), subsample=10, random=True, scale=0.001)
vis3d.pose(T_camera_world)
vis3d.pose(T_gripper_world_lift)
vis3d.pose(T_gripper_world)
vis3d.pose(T_cb_world)
vis3d.pose(RigidTransform())
vis3d.table(dim=0.5, T_table_world=T_cb_world)
vis3d.show()
if robot_type == 'ur5':
robot.movej(KINEMATIC_AVOIDANCE_JOINTS, wait_for_res=True)
robot.goto_pose(T_gripper_world_lift)
else:
robot.goto_pose(T_gripper_world_lift)
robot.goto_pose(T_gripper_world)
# wait for human measurement
yesno = raw_input('Take measurement. Hit [ENTER] when done')
robot.goto_pose(T_gripper_world_lift)
# choose target point 2
target_ind = np.where(ip == np.min(ip))[0]
target_pt_world = cb_points_world[target_ind[0]]
# create robot pose relative to target point
t_gripper_world = np.array([target_pt_world.x + config['gripper_offset_x'],
target_pt_world.y + config['gripper_offset_y'],
target_pt_world.z + config['gripper_offset_z']])
T_gripper_world = RigidTransform(rotation=R_gripper_world,
translation=t_gripper_world,
from_frame='gripper',
to_frame='cb')
logging.info('Moving robot to point x=%f, y=%f, z=%f' %(t_gripper_world[0], t_gripper_world[1], t_gripper_world[2]))
T_lift = RigidTransform(translation=(0,0,0.05), from_frame='cb', to_frame='cb')
T_gripper_world_lift = T_lift * T_gripper_world
robot.goto_pose(T_gripper_world_lift)
robot.goto_pose(T_gripper_world)
# wait for human measurement
yesno = raw_input('Take measurement. Hit [ENTER] when done')
robot.goto_pose(T_gripper_world_lift)
robot.goto_pose(T_orig_gripper_world_lift)
# choose target point 3
dir_world = np.array([1.0, 1.0, 0])
dir_world = dir_world / np.linalg.norm(dir_world)
ip = dir_world.dot(cb_point_data_world)
target_ind = np.where(ip == np.max(ip))[0]
target_pt_world = cb_points_world[target_ind[0]]
# create robot pose relative to target point
t_gripper_world = np.array([target_pt_world.x + config['gripper_offset_x'],
target_pt_world.y + config['gripper_offset_y'],
target_pt_world.z + config['gripper_offset_z']])
T_gripper_world = RigidTransform(rotation=R_gripper_world,
translation=t_gripper_world,
from_frame='gripper',
to_frame='cb')
logging.info('Moving robot to point x=%f, y=%f, z=%f' %(t_gripper_world[0], t_gripper_world[1], t_gripper_world[2]))
T_lift = RigidTransform(translation=(0,0,0.05), from_frame='cb', to_frame='cb')
T_gripper_world_lift = T_lift * T_gripper_world
robot.goto_pose(T_gripper_world_lift)
robot.goto_pose(T_gripper_world)
# wait for human measurement
yesno = raw_input('Take measurement. Hit [ENTER] when done')
robot.goto_pose(T_gripper_world_lift)
robot.goto_pose(T_orig_gripper_world_lift)
# stop robot
robot.reset_home()
if robot_type != 'ur5' and 'reset_bin' in config.keys() and config['reset_bin']:
y.reset_bin()
if robot_type == 'ur5':
robot.stop()
else:
y.stop()
sensor.stop()