-
Notifications
You must be signed in to change notification settings - Fork 0
/
AutoCalibration.lua
2143 lines (1869 loc) · 77.6 KB
/
AutoCalibration.lua
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
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
--[[
AutoCalibration.lua
Copyright (c) 2018, Xamla and/or its affiliates. All rights reserved.
This program is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License
as published by the Free Software Foundation; either version 2
of the License, or any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
--]]
local ros = require 'ros'
local tf = ros.tf
local actionlib = ros.actionlib
local SimpleClientGoalState = actionlib.SimpleClientGoalState
local xutils = require 'xamlamoveit.xutils'
local xamlamoveit = require 'xamlamoveit'
local datatypes = xamlamoveit.datatypes
local cv = require 'cv'
require 'cv.imgproc'
require 'cv.imgcodecs'
require 'cv.highgui'
require 'cv.calib3d'
require 'ximea.ros.XimeaClient'
local grippers = require 'xamlamoveit.grippers.env'
local gripper_force = 20
local prompt = xutils.Prompt()
-- Initialize the pseudo random number generator
math.randomseed( os.time() )
math.random(); math.random(); math.random()
local function tryRequire(module_name)
local ok, val = pcall(function() return require(module_name) end)
if ok then
return val
else
return nil
end
end
local xml = tryRequire('xml') -- for exporting the .t7 calibration file to .xml
local function readKeySpinning()
local function spin()
if not ros.ok() then
return false, 'ros shutdown requested'
else
ros.spinOnce()
return true
end
end
return xutils.waitKey(spin)
end
local autocal = require 'auto_calibration.env'
local CalibrationMode = autocal.CalibrationMode
local CalibrationFlags = autocal.CalibrationFlags
local AutoCalibration = torch.class('autoCalibration.AutoCalibration', autocal)
--creates a gripper client for the specified key
local function constructGripper(grippers, key, nh)
local gripper_action_name = nil
if string.find(key, 'robotiq') ~= nil then
gripper_force = 50
gripper_action_name = string.format('%s/gripper_command', key)
return grippers['GenericRosGripperClient'].new(nh, gripper_action_name)
elseif string.find(key, 'wsg') ~= nil then
local gripper_namespace = key
gripper_action_name = 'gripper_control'
--gripper_action_name = string.format('%s/gripper_control', key)
return grippers['WeissTwoFingerModel'].new(nh, gripper_namespace, gripper_action_name)
end
end
local function initializeGripperServices(self)
local node_handle = ros.NodeHandle()
self.node_handle = node_handle
local key = self.configuration.gripper_key -- get the key (gripper action name) from the configuration
self.gripper = constructGripper(grippers, key, self.node_handle)
if self.gripper ~= nil then
print('Gripper:')
print(self.gripper)
print('AutoCalibration calling home gripper')
self.gripper:home()
end
end
function alphanumsort(o)
local function conv(s)
local res, dot = "", ""
for n, m, c in tostring(s):gmatch"(0*(%d*))(.?)" do
if n == "" then
dot, c = "", dot..c
else
res = res..(dot == "" and ("%03d%s"):format(#m, m)
or "."..n)
dot, c = c:match"(%.?)(.*)"
end
res = res..c:gsub(".", "\0%0")
end
return res
end
table.sort(o,
function (a, b)
local ca, cb = conv(a), conv(b)
return ca < cb or ca == cb and a < b
end)
return o
end
function AutoCalibration:__init(configuration, move_group, camera_client, sl_studio)
self.configuration = configuration
self.config_class = ConfigurationCalibration.new(configuration)
self.move_group = move_group
self.camera_client = camera_client
self.sl_studio = sl_studio
if string.find(self.configuration.gripper_key, 'robotiq') then
gripper_force = 50
end
local ok, err = pcall(function() initializeGripperServices(self) end)
if not ok then
error('Gripper initialization failed: ' .. err)
end
self.tcp_frame_of_reference, self.tcp_end_effector_name = self:getEndEffectorName()
--create the calibrate/current folder if it does not exist
os.execute('mkdir -p ' .. configuration.output_directory)
--self.current_path = path.join(configuration.output_directory, 'current')
--os.execute('mkdir -p ' .. self.current_path)
-- create an output directory path so that jsposes.t7 and the calibration file are stored in the same directory
self.calibration_folder_name = os.date(configuration.calibration_directory_template)
self.output_directory = path.join(configuration.output_directory, self.calibration_folder_name)
print('Output directory for calibration data: '.. self.output_directory)
-- if we are simulating a capture, we will load an old jsposes file
self.offline_jsposes_fn = path.join(configuration.output_directory, 'offline', 'jsposes.t7')
end
function AutoCalibration:testMoveGroups()
local selected_move_group_name = self.configuration.move_group_name
local move_group_names, move_group_details = self.move_group.motion_service:queryAvailableMoveGroups()
print('AutoCalibration:testMoveGroups() move_group_names, move_group_details')
print(move_group_names)
print(move_group_details)
local index = 1
local tcp_frame_of_reference = move_group_details[selected_move_group_name].end_effector_link_names[index]
local tcp_end_effector_name = move_group_details[selected_move_group_name].end_effector_names[index]
print('config selected tcp_end_effector='..tcp_end_effector_name)
end
function AutoCalibration:getEndEffectorName()
local move_group_names, move_group_details = self.move_group.motion_service:queryAvailableMoveGroups()
local index = 1
local tcp_frame_of_reference = move_group_details[move_group_names[1]].end_effector_link_names[index]
local tcp_end_effector_name = move_group_details[move_group_names[1]].end_effector_names[index]
return tcp_frame_of_reference,tcp_end_effector_name
end
function AutoCalibration:shutdown()
if self.gripper ~= nil then
self.gripper:shutdown()
end
self.node_handle:shutdown()
end
local function createPatternLocalizer(self)
local pattern_geometry = self.configuration.circle_pattern_geometry
local pattern_localizer = autocal.PatternLocalisation()
pattern_localizer.circleFinderParams.minArea = 300
pattern_localizer.circleFinderParams.maxArea = 4000
pattern_localizer:setPatternIDdictionary(torch.load("/home/xamla/Rosvita.Control/lua/auto_calibration/patternIDdictionary.t7"))
pattern_localizer:setDBScanParams(100, 10)
pattern_localizer.debugParams = { circleSearch = false, clustering = false, pose = false }
pattern_localizer:setPatternData(pattern_geometry[2], pattern_geometry[1], pattern_geometry[3])
self.pattern_localizer = pattern_localizer
end
local function moveJ(self, pos)
assert(pos ~= nil, 'Target position is nil.')
self.move_group:moveJoints(pos, self.configuration.velocity_scaling)
--check that we arrived where we expected
local curPose = self.move_group:getCurrentPose():toTensor()
local curJoints = self.move_group:getCurrentJointValues()
end
local function moveJointsCollisionFree(end_effector, target_posture, velocity_scaling)
-- plan trajectory
local ok, err, joint_trajectory, ex_plan_parameters = pcall(function () return end_effector.move_group:planMoveJointsCollisionFree(target_posture, velocity_scaling) end)
if ok then
-- execute trajectory
local execute_ok = pcall(function () return end_effector.motion_service:executeJointTrajectory(joint_trajectory, true) end)
if execute_ok then
return true
else
print('Could not execute the joint trajectory.')
return false
end
else
print('Could not find a path to the given joint values.')
return false
end
end
local function movePoseCollisionFree(end_effector, target_pose, seed, velocity_scaling)
local plan_parameters = end_effector.move_group:buildPlanParameters(velocity_scaling)
local seed = seed or end_effector.move_group:getCurrentJointValues()
-- calculate joint values by inverse kinematic call
local ik_ok, solution =
end_effector.motion_service:queryIK(
target_pose,
plan_parameters,
seed:select(end_effector.move_group:getJointNames()),
end_effector.link_name
)
if ik_ok[1].val ~= 1 then
print('Failed inverse kinematic call')
return false
end
local goal = datatypes.JointValues(seed.joint_set:clone(), solution[1].values)
-- plan and execute trajectory
local success = moveJointsCollisionFree(end_effector, goal, velocity_scaling)
return success
end
function AutoCalibration:moveToStart()
local base_poses = self.configuration.base_poses
assert(base_poses ~= nil, 'Base poses are not defined.')
assert(base_poses['start'] ~= nil, 'Target position is nil.')
local ee_names = self.move_group:getEndEffectorNames()
local ee = self.move_group:getEndEffector(ee_names[1])
moveJointsCollisionFree(ee, base_poses['start'], 0.05)
end
function AutoCalibration:moveToCaptureBase()
local base_poses = self.configuration.base_poses
assert(base_poses ~= nil, 'Base poses are not defined.')
local ee_names = self.move_group:getEndEffectorNames()
local ee = self.move_group:getEndEffector(ee_names[1])
moveJointsCollisionFree(ee, base_poses['camera1_base'], self.configuration.velocity_scaling)
end
function AutoCalibration:pickCalibrationTarget()
self.gripper:home()
print('close the gripper')
self.gripper:move{width=0.0, speed=0.2, force=gripper_force, stop_on_block=false} -- move closed
local base_poses = self.configuration.base_poses
assert(base_poses ~= nil, 'Base poses are not defined.')
print('move to start')
moveJ(self, base_poses['start'])
print('open the gripper')
local t = self.gripper:move{width=0.05, speed=0.2, force=gripper_force} -- move open
print("Is the gripper open? Type 1 (Yes) or 2 (No) and press \'Enter\'.")
print("1 Yes")
print("2 No")
local open_check = io.read("*n")
if open_check ~= 1 then
return false
end
print('move to pre pick pose')
moveJ(self, base_poses['pre_pick_marker'])
print('move to pick pose')
moveJ(self, base_poses['pick_marker'])
print('grasp calibration target')
-- If width of calibration target may vary (i.e is assumed to be unknown),
-- set width to 0.0, otherwise width can be set to the width of our calibration pattern.
-- The "wsg50" needs the width of the part to be grasped -> set width to 0.0115.
t = self.gripper:grasp{width=0.0115, speed=0.2, force=gripper_force} -- grasp target
if t:hasCompletedSuccessfully() ~= true then
print("Grasp has not been successful.")
print("Want to try again? Type 1 (Yes) or 2 (No) and press \'Enter\'.")
local again = io.read("*n")
if again == 1 then
print('grasp calibration target')
t = self.gripper:grasp{width=0.0115, speed=0.2, force=gripper_force}
if t:hasCompletedSuccessfully() ~= true then
print("Again, grasping the pattern has not been successful.")
return false
end
else
return false
end
end
--t = self.gripper:move{width=0.0, speed=0.2, force=30, stop_on_block=false} -- move grasp
--assert(t:hasCompleted() == true, 'Cannot grasp pattern.')
print('move to post pick pose')
moveJ(self, base_poses['post_pick_marker'], self.configuration.velocity_scaling * 0.25)
print('move to start')
moveJ(self, base_poses['start'])
end
function AutoCalibration:returnCalibrationTarget()
local base_poses = self.configuration.base_poses
assert(base_poses ~= nil, 'Base poses are not defined.')
print('move to start')
moveJ(self, base_poses['start'])
print('move to post pick (= pre return) pose')
moveJ(self, base_poses['post_pick_marker'])
print('move to pick (= return) pose')
sys.sleep(3)
moveJ(self, base_poses['pick_marker'], self.configuration.velocity_scaling * 0.25)
print('release calibration target')
local t = self.gripper:release{width=0.05, speed=0.2, force=gripper_force} -- release target
if t:hasCompletedSuccessfully() ~= true then
print("Failed to release the pattern.")
print("Want to try again? Type 1 (Yes) or 2 (No) and press \'Enter\'.")
local again = io.read("*n")
if again == 1 then
print('release calibration target')
t = self.gripper:release{width=0.05, speed=0.2, force=gripper_force}
if t:hasCompletedSuccessfully() ~= true then
print("Again, failed to release the pattern.")
return false
end
else
return false
end
end
print('move to pre pick (= post return) pose')
moveJ(self, base_poses['pre_pick_marker'])
end
function AutoCalibration:simulateCapture()
-- assume the captured images are already at the capture folder
--load jsposes.t7 file from the offline folder
local offline_jsposes
if path.exists(self.offline_jsposes_fn) then
print('Reading offline poses file ' .. self.offline_jsposes_fn)
offline_jsposes = torch.load(self.offline_jsposes_fn)
else
print(self.offline_jsposes_fn.. ': file does not exist')
return false
end
-- we restore the values so they can be saved afterwards
self.recorded_joint_values = {}
self.recorded_poses = {}
for i = 1, #offline_jsposes.recorded_joint_values do
print('restoring pose #'..i)
self.recorded_joint_values[i] = offline_jsposes.recorded_joint_values[i]
self.recorded_poses[i] = offline_jsposes.recorded_poses[i]
end
self:savePoses()
end
function AutoCalibration:runCaptureSequenceWithoutCapture()
local pos_list = self.configuration.capture_poses
assert(pos_list ~= nil and #pos_list > 0)
local recorded_joint_values = {} -- joint values
local recorded_poses = {} -- end effector poses
for i,p in ipairs(pos_list) do
printf('Moving to position #%d...', i)
local ee_names = self.move_group:getEndEffectorNames()
local ee = self.move_group:getEndEffector(ee_names[1])
moveJointsCollisionFree(ee, p, self.configuration.velocity_scaling)
sys.sleep(1)
-- get joint values and pose
local joint_values = self.move_group:getCurrentJointValues()
local pose = self.move_group:getCurrentPose()
recorded_joint_values[#recorded_joint_values+1] = joint_values
recorded_poses[#recorded_poses+1] = pose:toTensor()
end
self.recorded_joint_values = recorded_joint_values
self.recorded_poses = recorded_poses
self:savePoses()
end
function AutoCalibration:closeGripper()
if self.gripper ~= nil then
print('close the gripper')
self.gripper:move{width=0.0, speed=0.2, force=gripper_force, stop_on_block=false} -- move closed
else
print('Need to initialize the gripper')
end
end
function AutoCalibration:openGripper()
if self.gripper ~= nil then
print('open the gripper')
self.gripper:move{width=0.06, speed=0.2, force=gripper_force} -- move open
else
print('Need to initialize the gripper')
end
end
function AutoCalibration:homeGripper()
if self.gripper ~= nil then
print('Homeing gripper')
self.gripper:home()
else
print('Need to initialize the gripper')
end
end
local function captureImage(self, i, camera_configuration, output_directory)
local camera_serial = camera_configuration.serial
local exposure = camera_configuration.exposure
local sleep_before_capture = camera_configuration.sleep_before_capture
-- wait configured time before capture
if sleep_before_capture > 0 then
printf('wait before capture %f s... ', sleep_before_capture)
sys.sleep(sleep_before_capture)
end
-- capture image
self.camera_client:setExposure(exposure, {camera_serial})
local image = self.camera_client:getImages({camera_serial})
-- get joint values and pose of image
local joint_values = self.move_group:getCurrentJointValues()
local pose = self.move_group:getCurrentPose()
-- create output filename
--output_directory = self.config_class:getCameraDataOutputPath(camera_serial)
local fn = path.join(output_directory, string.format('cam_%s_%03d.png', camera_serial, i))
if image:nDimension() > 2 then
image = cv.cvtColor{image, nil, cv.COLOR_RGB2BGR}
end
-- write image to disk
printf("Writing image: %s", fn)
local ok = cv.imwrite{fn, image}
assert(ok, 'Could not write image.')
return fn, joint_values, pose
end
local function tryLoadCurrentCameraCalibration(self, camera_serial)
--local current_output_directory = path.join(self.configuration.output_directory, 'current')
local calibration_fn = string.format('cam_%s.t7', camera_serial)
--local calibration_file_path = path.join(current_output_directory, calibration_fn)
local calibration_file_path = path.join(self.output_directory, calibration_fn)
printf("Probing for calibration file '%s'.", calibration_file_path)
if path.exists(calibration_file_path) then
return torch.load(calibration_file_path)
else
return nil
end
end
local function tryLoadCurrentCameraCalibrationFromStereo(self, camera_left_serial, camera_right_serial)
--local current_output_directory = path.join(self.configuration.output_directory, 'current')
local calibration_fn = string.format('stereo_cams_%s_%s.t7', camera_left_serial, camera_right_serial)
--local calibration_file_path = path.join(current_output_directory, calibration_fn)
local calibration_file_path = path.join(self.output_directory, calibration_fn)
printf("Probing for calibration file '%s'.", calibration_file_path)
if path.exists(calibration_file_path) then
return torch.load(calibration_file_path)
else
return nil
end
end
function getHomogeneousFromRosStampedPose(msg)
--convert to tensor
transf = tf.Transform.new()
q = msg.pose.orientation
t = msg.pose.position
transf:setRotation(tf.Quaternion.new(q.x,q.y,q.z,q.w))
transf:setOrigin({t.x,t.y,t.z})
return transf:toTensor()
end
function AutoCalibration:runCaptureSequence()
local configuration = self.configuration
local file_names = {}
local generic_file_names = {}
local recorded_joint_values = {} -- joint values after getImage calls
local recorded_poses = {} -- end effector poses after getImage calls
local output_directory = path.join(configuration.output_directory, 'capture')
print('Deleting output directory')
os.execute('rm -rf '.. output_directory)
print('Creating output directory')
os.execute('mkdir -p ' .. output_directory)
local pos_list = configuration.capture_poses
assert(pos_list ~= nil and #pos_list > 0)
local left_camera = configuration.cameras[configuration.left_camera_id]
local right_camera = configuration.cameras[configuration.right_camera_id]
for i,p in ipairs(pos_list) do
printf('Moving to position #%d...', i)
-- move to capture pose
local ee_names = self.move_group:getEndEffectorNames()
local ee = self.move_group:getEndEffector(ee_names[1])
moveJointsCollisionFree(ee, p, self.configuration.velocity_scaling)
if left_camera ~= nil then
local fn, joint_values, pose = captureImage(self, i, left_camera, output_directory)
file_names[#file_names+1] = fn
recorded_joint_values[#recorded_joint_values+1] = joint_values
recorded_poses[#recorded_poses+1] = pose:toTensor()
end
if right_camera ~= nil then
local fn, joint_values, pose = captureImage(self, i, right_camera, output_directory)
file_names[#file_names+1] = fn
end
collectgarbage()
end
self.file_names = file_names
self.recorded_joint_values = recorded_joint_values
self.recorded_poses = recorded_poses
self:savePoses()
return file_names, recorded_joint_values, recorded_poses
end
local function CalcPointPositions (arg)
-- Calculate the "true" 3D position (x,y,z) of the circle centers of the circle pattern.
-- z position is set to 0 for all points
--
-- Input params:
-- arg.pointsX -- number of points in horizontal direction
-- arg.pointsY -- number of points in vertical direction
-- arg.pointDistance -- distance between two points of the pattern in meter
-- Return value:
-- Position of the circle centers
local corners = torch.FloatTensor(arg.pointsX*arg.pointsY, 1, 3):zero()
local i=1
for y=1, arg.pointsY do
for x=1, arg.pointsX do
corners[i][1][1] = (2*(x-1) + (y-1)%2) * arg.pointDistance
corners[i][1][2] = (y-1)*arg.pointDistance
corners[i][1][3] = 0
i = i+1
end
end
return corners
end
local function normalize(v)
return v / torch.norm(v)
end
-- Calculates the TCP pose required to point the TCP z axis to point at with TCP at position eye
-- eye becomes origin, 'at' lies on x-axis
-- @param eye torch.Tensor(3,1); TCP position in x, y, z
-- @param at torch.Tensor(3,1); Point to look at
-- @param up torch.Tensor({0, 0, 1}); up direction of the camera
-- @return torch.Tensor(4,4), robot pose
local function PointAtPose(eye, at, up, handEye)
local zaxis = normalize(at - eye)
local xaxis = normalize(torch.cross(zaxis, up))
local yaxis = torch.cross(zaxis, xaxis)
local basis = torch.Tensor(3,3)
basis[{{},{1}}] = xaxis
basis[{{},{2}}] = yaxis
basis[{{},{3}}] = zaxis
local t = tf.Transform()
t:setBasis(basis)
t:setOrigin(eye)
local tcpLookat=t:toTensor():clone()
if handEye ~= nil then
tcpLookat = tcpLookat * torch.inverse(handEye)
end
return tcpLookat
end
local function RotVectorToRotMatrix(vec)
-- transform a rotation vector as e.g. provided by solvePnP to a 3x3 rotation matrix using the Rodrigues' rotation formula
-- see e.g. http://docs.opencv.org/2.4/modules/calib3d/doc/camera_calibration_and_3d_reconstruction.html#void%20Rodrigues%28InputArray%20src,%20OutputArray%20dst,%20OutputArray%20jacobian%29
--
-- Input parameters:
-- vec = vector to transform
-- Return value:
-- 3x3 rotation matrix
local theta = torch.norm(vec)
local r = vec/theta
r = torch.squeeze(r)
local mat = torch.Tensor({{0, -1*r[3], r[2]}, {r[3], 0, -1*r[1]}, {-1*r[2], r[1], 0}})
r = r:resize(3,1)
local result = torch.eye(3)*math.cos(theta) + (r*r:t())*(1-math.cos(theta)) + mat*math.sin(theta)
return result
end
local function transformMatrixToQuaternion(rot)
local sqrt = math.sqrt
local trace = rot[1][1] + rot[2][2] + rot[3][3]
local _next = { 2, 3, 1 }
local q = torch.zeros(4)
if trace > 0 then
local r = sqrt(trace + 1)
local s = 0.5 / r
q[1] = 0.5 * r
q[2] = (rot[3][2] - rot[2][3]) * s
q[3] = (rot[1][3] - rot[3][1]) * s
q[4] = (rot[2][1] - rot[1][2]) * s
else
local i = 1
if rot[2][2] > rot[1][1] then
i = 2
end
if rot[3][3] > rot[i][i] then
i = 3
end
local j = _next[i]
local k = _next[j]
local t = rot[i][i] - rot[j][j] - rot[k][k] + 1
local r = sqrt(t)
local s = 0.5 / sqrt(t)
local w = (rot[k][j] - rot[j][k]) * s
q[1] = w
q[i+1] = 0.5 * r
q[j+1] = (rot[j][i] + rot[i][j]) * s
q[k+1] = (rot[k][i] + rot[i][k]) * s
end
return q/q:norm()
end
local function transformQuaternionToMatrix(q)
local w = q[1]
local x = q[2]
local y = q[3]
local z = q[4]
local result = torch.DoubleTensor(3,3)
result[1][1] = 1 - 2*y*y - 2*z*z
result[1][2] = 2*x*y - 2*w*z
result[1][3] = 2*x*z + 2*w*y
result[2][1] = 2*x*y + 2*w*z
result[2][2] = 1 - 2*x*x - 2*z*z
result[2][3] = 2*y*z - 2*w*x
result[3][1] = 2*x*z - 2*w*y
result[3][2] = 2*y*z + 2*w*x
result[3][3] = 1 - 2*x*x - 2*y*y
return result
end
local function calc_avg_heye(H1, H2)
local Q = torch.DoubleTensor(2, 4)
local avg_pos = torch.zeros(3)
avg_pos = (H1[{{1,3},{4}}] + H2[{{1,3},{4}}]) / 2.0
Q[1] = transformMatrixToQuaternion(H1[{{1,3},{1,3}}])
Q[2] = transformMatrixToQuaternion(H2[{{1,3},{1,3}}])
local QtQ = Q:t() * Q
local e, V = torch.symeig(QtQ, 'V')
local maxEigenvalue, maxEig_index = torch.max(e,1)
local avg_q = V:t()[maxEig_index[1]]
local avg_rot = transformQuaternionToMatrix(avg_q)
local avg_H = torch.DoubleTensor(4,4)
avg_H[{{1,3},{1,3}}] = avg_rot
avg_H[{{1,3},{4}}] = avg_pos
avg_H[4][4] = 1.0
return avg_H
end
-- Sphere sampling with onboard camera setup
function AutoCalibration:captureSphereSampling_endOfArmCams()
attr = xutils.saveTerminalAttributes()
prompt:enableRawTerminal()
print('How many? Enter the number of capture poses:')
local count = prompt:readNumber()
local min_radius = 0.35
local max_radius = 0.6
local target_jitter = 0.015
print("Enter filename of initial guess for hand-eye matrix (without quotation marks!)")
local heye_fn = prompt:readLine()
local heye = nil
if heye_fn ~= "" then
heye = torch.load(heye_fn)
else
print("Hand-Eye-Matrix not found!")
return nil
end
print("Hand-eye matrix:")
print(heye)
print("Note: If hand-eye matrix is for tcp at last joint of robot arm (instead of tcp at gripper tip), then the tcp in the project configuration has to be set accordingly before starting the sphere sampling!")
print("Enter filename of initial guess for stereo calibration")
local stereoCalib_fn = prompt:readLine()
local intrinsics = nil
local distortion = nil
local interCamTrafo = nil
if stereoCalib_fn ~= "" then
stereoCalib = torch.load(stereoCalib_fn)
intrinsics = stereoCalib.camLeftMatrix:double()
distortion = stereoCalib.camLeftDistCoeffs:double()
interCamTrafo = stereoCalib.trafoLeftToRightCam:double()
else
print("Stereocalibration not found!")
return nil
end
print("Left camera matrix:")
print(intrinsics)
print("Left camera distortion:")
print(distortion)
print("Transformation between cameras:")
print(interCamTrafo)
local heye_2 = heye * torch.inverse(interCamTrafo)
print("Hand-eye matrix of right camera:")
print(heye_2)
local avg_heye = calc_avg_heye(heye, heye_2)
print("average hand-eye:")
print(avg_heye)
local ok, centers = false, nil
local pos_list = {}
local file_names = {}
local recorded_joint_values = {} -- joint values after getImage calls
local recorded_joint_tensors = {} -- joint values as tensor after getImage calls
local recorded_poses = {} -- end effector poses after getImage calls
local output_directory = path.join(self.configuration.output_directory, 'capture_sphere_sampling')
print('Deleting output directory')
os.execute('rm -rf '.. output_directory)
print('Creating output directory')
os.execute('mkdir -p ' .. output_directory)
local left_camera = self.configuration.cameras[self.configuration.left_camera_id]
local right_camera = self.configuration.cameras[self.configuration.right_camera_id]
local ee_names = self.move_group:getEndEffectorNames()
local ee = self.move_group:getEndEffector(ee_names[1])
local overviewPose = self.move_group:getCurrentPose()
local check = false
while ros.ok() and not ok do
print("Please move robot to overview pose and press 'Enter' when ready.")
print("Cameras have to look straight down to the table and capture the whole pattern.")
prompt:waitEnterOrEsc()
overviewPoseTensor = self.move_group:getCurrentPose():toTensor()
print("overviewPoseTensor:")
print(overviewPoseTensor)
-- capture images
local image_left
local image_right
if left_camera ~= nil then
if left_camera.sleep_before_capture > 0 then
printf('wait before capture %f s... ', left_camera.sleep_before_capture)
sys.sleep(left_camera.sleep_before_capture)
end
camera_client:setExposure(left_camera.exposure, {left_camera.serial})
image_left = camera_client:getImages({left_camera.serial})
if image_left:nDimension() > 2 then
image_left = cv.cvtColor{image_left, nil, cv.COLOR_RGB2BGR}
end
end
if right_camera ~= nil then
if right_camera.sleep_before_capture > 0 then
printf('wait before capture %f s... ', right_camera.sleep_before_capture)
sys.sleep(right_camera.sleep_before_capture)
end
camera_client:setExposure(right_camera.exposure, {right_camera.serial})
image_right = camera_client:getImages({right_camera.serial})
if image_right:nDimension() > 2 then
image_right = cv.cvtColor{image_right, nil, cv.COLOR_RGB2BGR}
end
end
-- write images to disk
fn_startImg_left = string.format('cam_%s_start.png', left_camera.serial)
printf("Writing image: %s", fn_startImg_left)
ok_write_start_left = cv.imwrite{ fn_startImg_left, image_left }
assert(ok_write_start_left, 'Could not write left start image.')
fn_startImg_right = string.format('cam_%s_start.png', right_camera.serial)
printf("Writing image: %s", fn_startImg_right)
ok_write_start_right = cv.imwrite{ fn_startImg_right, image_right }
assert(ok_write_start_right, 'Could not write right start image.')
-- load images for simulation
--image_left = cv.imread { string.format('cam_%s_start.png', left_camera.serial) }
--image_right = cv.imread { string.format('cam_%s_start.png', right_camera.serial) }
ok_left, centers_left = cv.findCirclesGrid{ image = image_left,
patternSize = { height = self.configuration.circle_pattern_geometry[1], width = self.configuration.circle_pattern_geometry[2] },
flags=cv.CALIB_CB_ASYMMETRIC_GRID + cv.CALIB_CB_CLUSTERING }
local ok_right = true -- if we do not have a right camera, ok_right should be true per default
if right_camera ~= nil then
ok_right, centers_right = cv.findCirclesGrid{ image = image_right,
patternSize = { height = self.configuration.circle_pattern_geometry[1], width = self.configuration.circle_pattern_geometry[2] },
flags=cv.CALIB_CB_ASYMMETRIC_GRID + cv.CALIB_CB_CLUSTERING }
end
ok = ok_left and ok_right
print("ok:")
print(ok)
if not ok then
print('WARNING! Calibration pattern not found. Please move the target into camera view!')
end
end
local circlePositions = CalcPointPositions{ pointsX = self.configuration.circle_pattern_geometry[2],
pointsY = self.configuration.circle_pattern_geometry[1],
pointDistance = self.configuration.circle_pattern_geometry[3] }
local poseFound, poseCamRotVector, poseCamTrans = cv.solvePnP{ objectPoints = circlePositions,
imagePoints = centers_left,
cameraMatrix = intrinsics,
distCoeffs = distortion }
if not poseFound then
error('could not calculate pose from calibration pattern')
end
print("poseCamRotVector:")
print(poseCamRotVector)
print("poseCamTrans:")
print(poseCamTrans)
local poseCamRotMatrix = RotVectorToRotMatrix(poseCamRotVector)
print("poseCamRotMatrix:")
print(poseCamRotMatrix)
-- assemble the 4x4 transformation matrix
local transfer = torch.eye(4)
transfer[{{1,3}, {1,3}}] = poseCamRotMatrix
transfer[{{1,3}, {4}}] = poseCamTrans
print("transfer:")
print(transfer)
local offset = torch.mv(transfer, torch.Tensor({0.04,0.05,0,0}))
transfer[{{},4}]:add(offset)
local t = overviewPoseTensor * avg_heye * transfer
targetPoint = t[{{1,3},4}]
print('identified target point:')
print(targetPoint)
prompt:anyKey()
-- Once write all joint values to disk (e.g. to get the torso joint value of an SDA, etc.)
local mg_names = self.move_group.motion_service:queryAvailableMoveGroups()
local mg_all = self.move_group.motion_service:getMoveGroup(mg_names[1])
local joint_values_all = mg_all:getCurrentJointValues()
local all_vals_fn1 = path.join(output_directory, "all_vals.t7")
local all_vals_fn2 = path.join(output_directory, "all_vals_tensors.t7")
torch.save(all_vals_fn1, joint_values_all)
torch.save(all_vals_fn2, joint_values_all.values)
-- Preparation of saving joint values and poses to disk after each move
local jsposes = {}
jsposes.recorded_joint_values = {}
jsposes.recorded_poses = {}
local jsposes_tensors = {}
jsposes_tensors.recorded_joint_values = {}
jsposes_tensors.recorded_poses = {}
local poses_fn1 = path.join(output_directory, "jsposes.t7")
local poses_fn2 = path.join(output_directory, "jsposes_tensors.t7")
-- Loop over all #count many poses, the user wants to sample
local up = torch.DoubleTensor({0,0, 1})
local cnt = 1
local enough = false
while not enough do
-- generate random point in positive half shere
local origin
while true do
origin = torch.randn(3) -- 1D Tensor of size 3 filled with random numbers
-- from a normal distribution with mean zero and variance one.
origin[3] = math.max(0.01, math.abs(origin[3])) -- z-component has to be positive
origin:div(origin:norm()) -- normalization
if origin[3] > 0.8 then -- z-component has to be > 0.8
break
end
end
print("origin:")
print(origin)
-- Lets express the position we want to look at relative to our pattern.
-- The targets z-axis goes into the table so we have a negative z-value w.r.t. the pattern.
origin:mul(torch.lerp(min_radius, max_radius, math.random())) -- = minRadius + random in [0,1) * (maxRadius-minRadius)
-- ==> Each point has to lie on a sphere with radius between minRadius and maxRadius!
print("origin:")
print(origin)
origin:add(targetPoint)
print("origin:")
print(origin)
local target = targetPoint + math.random() * target_jitter - 0.5 * target_jitter
local up_ = up
up_ = t[{1,{1,3}}] -- use pattern x axis in world
-- with 50% random choice rotate 180° around z-axis
if math.random(2) == 1 then
up_ = -up_
end
print("up_:")
print(up_)
local movePoseTensor = PointAtPose(origin, target, up_, avg_heye)
print("movePoseTensor:")
print(movePoseTensor)
local movePose = datatypes.Pose()
movePose.stampedTransform:fromTensor(movePoseTensor)
movePose:setFrame("world")
print("movePose:")
print(movePose)
-- move to movePose and search calibration pattern
print("current pose:")
print(self.move_group:getCurrentPose())
printf('Moveing to movePose #%d ...', cnt)
-- with this seed being set to random values, we have an arm rebuilding
local seed = ee.move_group:getCurrentJointValues()
seed.values = torch.randn(seed.values:size(1)) -- 1D Tensor of same size as seed filled with random numbers in ]-1,+1[
check = movePoseCollisionFree(ee, movePose, seed, self.configuration.velocity_scaling)
print("check:")
print(check)
if check then
sys.sleep(0.5)
--prompt:anyKey()
-- Capture images and save joint values and poses:
if left_camera ~= nil then
camera_client:setExposure(left_camera.exposure, {left_camera.serial})
local image = camera_client:getImages({left_camera.serial})
if image:nDimension() > 2 then
image = cv.cvtColor{image, nil, cv.COLOR_RGB2BGR}
end
-- write image to disk
local fn = path.join(output_directory, string.format('cam_%s_%03d.png', left_camera.serial, cnt))
printf("Writing image: %s", fn)
local ok_write = cv.imwrite{fn, image}
assert(ok_write, 'Could not write image.')
file_names[#file_names+1] = fn
end
if right_camera ~= nil then
camera_client:setExposure(right_camera.exposure, {right_camera.serial})
local image = camera_client:getImages({right_camera.serial})
if image:nDimension() > 2 then
image = cv.cvtColor{image, nil, cv.COLOR_RGB2BGR}
end
-- write image to disk