Professional Documents
Culture Documents
NOVEMBER 2009
3
4
5
Laser beam welding bears evident advantages regarding precision, quality, productivity, low heat
input, and feasibility of automation. At the same time the process calls for high precision of the
beam positioning on the workpiece, which therefore imposes high requirements of welding
trajectory and feed rate accuracy; e.g., for butt welding the focal point of the laser beam with respect
to the joint must be maintained within a typical accuracy better than 20 150 m, depending on the
focused beam radius. To meet these requirements, seam-tracking devices are used. A sensor
measures the joint position and computes a correction vector to compensate the joint trajectory
offset. The deviation is compensated either by a robot trajectory adjustment or by an additional
tracking axis. This paper describes the basic concepts of seam tracking in detail and points out
problems in the different control principles, which are evoked by the forerun of the sensor.
State-of-the-art sensors and error compensating techniques are presented and analyzed. Further, a
new approach for seam tracking is introduced. It uses a multisensor concept, which in addition to the
seam position measures the relative displacement between the processing head and the workpiece.
An integrated two-dimensional beam positioning system enables self-guided processing, which
allows high-accuracy tracking of a joint independent of the motion system and disengages from time
intensive sensor calibration and robot teaching necessity. 2009 Laser Institute of America.
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23 I. INTRODUCTION
1042-346X/2009/214/1/0/$25.00
algorithms, and mechanical improvements;2,57 however, articulated robots are not applicable for most butt- or filletwelding applications.
Comparable challenging to the robot path accuracy is the
compliance of the joint position accuracy, imposing high demands on workpiece accuracy and fixture repeatability, in
particular considering workpiece distortion through heat input of the welding process. This is often the reason for process irregularities and weld defects.8
In order to reduce the accuracy requirements of the
workpiece geometry and the fixture and therefore to enable
laser butt welding for mass production industry, seamtracking devices are used. As explained in Sec. III C, standard seam-tracking devices only compensate for joint trajectory of fixture deviations andafter calibrationrepetitious
path deviations of the robot. Deviations in robot repeatability
or thermal movement of the workpiece or fixture can only be
corrected by seam-tracking systems without sensor forerun9
or with a relative movement feedback as described in Sec.
IV D that closes the control loop around the end effector
position.
49
70
71
72
73
74
75
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
79 A. Sensor concept
103
108
127
128
129
130
131
132
133
134
135
136
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
137
138
139
140
141
142
143
144
145
direction of travel
joint
sensor
TCP
welding
head
corr
tracking
axis
robot
hand
TCP
x
y
146
joint
robot
robot arm
FIG. 6. Seam-tracking setup with sensor fixed to the TCP closed loop
configuration.
robot hand. The welding head with the TCP tool center
point; the point of incident of the welding laser sits on a
tracking axis, which allows adjusting the TCP relative to the
robot hand Fig. 5. The tracking axis is usually a precise
and fast linear axis, but can also be a rotary axis.23
In a more common setup the sensor is fixed to the
welding head Figs. 6 and 7. The advantage of this setup is
that the sensor moves with the TCP. Since the TCP matches
the current joint position, the sensor is continuously
readjusted and therefore covers a greater deviation between
joint and robot trajectory.24 The tracking axis may be
FIG. 5. Seam-tracking setup with sensor fixed to the robot hand open loop
configuration.
166
167
168
169
170
171
172
173
174
175
176
4 217
5 220
The nominal position of the tracking axis pcorr,y equals the 221
difference of the target position rtarget,y and the robot 222
position rrobot,y. With formulas 3 and 5 it can be 223
224
calculated to
194
195
196
197
s = rjoint rTCP ,
225
226
6 227
228
229
230
231
232
7 233
234
235
255
256
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
260
261
FIG. 9. Simplified seam-tracking control principle for closed loop configuration, not considering rotation and lateral robot movement.
262
292
293
294
295
276
273
274
275
277
278
279
280
281
282
283
284
rrobot,x,real
,
rrobot,x,exp
297
305
306
307
308
309
310
10 311
11 312
with
R=
FIG. 12. Deviation of the TCP position rotation caused by welding head
rotation .
313
cos sin
sin
cos
12
314
320
321
322
4: Workpiece
movement
13
14
types is to abridge the sensor offset sx. A significant shortening is realized with a coaxial sensor setup, e.g., Refs. 23, 25,
and 29.
For fillet or flange welds, Jackel et al.21 use the filler
wire instead of an optical sensor for seam tracking. The filler
wire receives a force if abutted on the joint edge, which is
gauged to control the correction axis. Since the tip of the
filler wire is coincident with the laser TCP, this principle
possesses zero sensor forerun, avoiding all previous described seam-tracking errors.
If a sensor forerun cannot be avoided, production capable seam-tracking sensors use different correction tech-
tracking
axis
paxis , y
(no correction)
rTCP , y (rTCP , x )
s y (rTCP , x + s x )
r jo int, y (rTCP , x + s x )
rrobot , y
All errors described in the previous section are originated in the sensor forerun with respect to the TCP. Therefore, a way to reduce sensor-originated deviation in all error
joint
trajectory
robot
scal,y
335
336
337
3: Rotation of
welding head
2: Lateral robot
movement
319
Reference workpiece
with ideal contour
318
D: Self-guided
processing
Corrected error
1: Variation of feed
rate
C: Two-sensor position
measurements
Robot-sensor
calibration/sync
317
B: Calibration on
reference workpiece
(ideal contour)
tracking
sensor
Calibration vector
(stored for work run)
FIG. 14. Storing of the calibration vector Scalt using a reference workpiece
with ideal joint trajectory rjoint,y = 0.
338
339
340
341
342
343
344
345
346
347
348
349
362
363
350
365
366
367
t=tT
364
rrobot,xt = sx .
15
368
369
370
371
373
374
375
376
377
378
379
380
381
16 382
FIG. 16. Correction of error type 2 using a second sensor s2 with a different forerun distance s2,x s1,x.
372
383
384
385
386
387
388
397
398
399
17
with
Ts = s2,x s1,x/rrobot,x .
Knowing both the TCP and the joint trajectory, the lateral
401 TCP position rTCP,y can be set onto the corresponding
402 lateral joint position Fig. 16. This correction technique
403 requires a known or constant feed rate and no welding
404 head rotation.
rTCPt = rconst +
With this concept, seam tracking is performed twodimensional position based instead of time based. All four
error types described in Sec. IV are inexistent.
The concept can be enhanced by a feedback of the target
position to the robot control dotted line in Fig. 17. With
this feedback, seam tracking of unknown joint trajectories is
feasible. In this case, the high accuracy positioning is
performed by the two dimensional tracking axis, while the
rough contour is followed by the robot.
437
446
447
A. Displacement sensor
448
449
450
joint
trajectory
robot
r
rrobot
tracking
axis
r
paxis
+
+
19
r
r joint
r
rTCP
displacement
sensor
r
pcorr
r TCPt dt,
431
432
433
434
435
20
Knowing both the TCP and the joint trajectory, the lateral
TCP position rTCP,y can be set onto the corresponding
lateral joint position,
rtargett = rjointirjoint,xi=rTCP,xt .
The correction vector equals the vector difference,
21
r
s joint
rtarget
r
r joint
r
r
rtarget (t ) = r joint (i )
tracking
sensor
r&TCP
430
22 436
438
439
440
441
442
443
444
445
18
400
429
rTCP
r joint , x (i )= rTCP , x (t )
FIG. 17. Correction of all four error types using an additional sensor to track
the relative displacement to the workpiece self-guided processing.
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
10
FIG. 18. Measurement of relative displacement between consecutive images by finding the maximum cross correlation of areas in consecutive observed
images.
480 C. Results
D. Further development
492
493
VI. CONCLUSION
511
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
11
FIG. 19. Optical setup of the combined displacement and seam-tracking sensor left: stand-alone sensor and right: coaxial setup.
514
FIG. 20. Observed workpiece surface and welding process coaxial setup.
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
12
FIG. 22. Coaxial setup of camera sensor, illumination, and laser positioning
system.
543
A concept is introduced, which combines a coaxial measurement concept with a laser positioning system. A processing head using this concept is capable to follow joints or
546 traces on the workpiece self-guided, which means indepen547 dent of the actual movement of the processing head and
548 without any prior calibration or information of the trajectory.
549 The absolute position accuracy is independent from the ac550 curacy of the motion system, which enables applications
551 with low accuracy robots or even hand moved systems.
544
545
552 Nomenclature
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
25
13
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687