search

JorgeFernandes-Git / zau_bot

4 stars 1 forks source link

e6 - Full calibration (2 RGB cameras and one 3D lidar) #13

Open JorgeFernandes-Git opened 1 year ago

JorgeFernandes-Git commented 1 year ago

Summary:

e6_multi_modal

Calibration tree and Transformations:

summary

e6_mm_calibration

Videos:

Calibration Results per collection:

Collection camera (px) camera_mb (px) velodyne (m)
000 0.5561 0.5320 0.0064
001 0.5263 0.5386 0.0064
002 0.5308 0.5346 0.0065
003 0.4642 0.5373 0.0060
004 0.6165 0.5390 0.0066
005 0.4792 0.5398 0.0062
007 0.5577 0.7607 0.0064
008 0.5560 0.6584 0.0062
009 1.0303 0.6434 0.0062
010 0.7677 0.6300 0.0067
011 0.5588 0.6184 0.0064
012 0.4670 0.6080 0.0064
013 0.2949 0.4613 0.0064
014 0.3067 0.4478 0.0067
015 0.2968 0.2064 0.0063
016 4.0170 3.8441 0.0088
017 3.0238 3.0225 0.0086
018 0.4867 0.5048 0.0064
019 0.4108 0.6225 0.0069
020 0.3086 0.4979 0.0064
021 0.3161 0.4620 0.0062
022 0.3217 0.4307 0.0069
023 0.8256 0.4673 0.0062
024 0.4339 0.4116 0.0061
025 0.7071 0.6554 0.0065
026 0.5492 0.9245 0.0063
027 0.6525 1.0532 0.0062
028 0.9089 1.1664 0.0061
029 0.7902 1.1462 0.0063
030 0.7275 1.0891 0.0065
Averages 0.7496 0.8185 0.0065

Results from the simulation using noise initial guess nig 0.1 0.1

Ground truth frame evaluations using different datasets

rosrun atom_evaluation ground_truth_frame_evaluation -train_json  /home/jorge/datasets/e6_mm/atom_calibration_nig_0.1_0.1.json -test_json /home/jorge/datasets/e6_mm/dataset_corrected.json
frame # Xcal-Xgt (mm) Ycal-Ygt (mm) Zcal-Zgt (mm) Roll_cal-Roll_gt (deg) Pitch_cal-Pitch_gt (deg) Yaw_cal-Yaw_gt (deg) Average - Trans Average - Rot
base_link_mb_to_base_link 9.2194 9.1957 17.4085 0.0491 0.0806 0.0806 21.7397 0.1924
camera 7.0570 20.7202 0.9205 0.0659 0.0101 0.0101 16.6333 0.3626
camera_mb 0.6239 22.4860 23.8324 0.0559 0.2354 0.2354 32.7719 0.4130
velodyne 1.7709 4.6873 26.4831 0.2674 0.0021 0.0021 26.9529 0.2849

Lidar to AGV camera evaluation (same dataset - Atom_calibration)

rosrun atom_evaluation lidar_to_rgb_evaluation -train_json /home/jorge/datasets/e6_mm/atom_calibration_nig_0.1_0.1.json -test_json /home/jorge/datasets/e6_mm/atom_calibration_nig_0.1_0.1.json -cs camera_mb -rs velodyne -si
Collection # RMS (pix) X err (pix) Y err (pix)
000 5.0165 3.1572 1.4601
001 5.5321 3.7744 1.4089
002 5.4894 3.9398 1.4221
003 3.6184 2.5606 1.0372
004 4.6210 3.5103 1.3770
005 2.6547 1.7478 0.7602
007 5.8273 3.5149 1.3806
008 4.9740 3.2375 1.3606
009 4.3480 2.8449 1.4139
010 6.3855 4.2268 1.5991
011 4.4527 3.0954 1.4694
012 4.8573 3.4109 1.4984
013 4.1894 2.8002 1.5317
014 4.3140 3.0491 1.4918
015 4.9254 3.1945 1.5025
016 2.1563 1.4797 0.5857
017 14.7812 3.3613 2.8118
018 22.8064 4.7569 5.0110
019 13.5416 4.4927 3.0352
020 4.2044 2.8963 1.3374
021 5.0278 3.3680 1.6414
022 14.8617 4.3641 4.0795
023 4.6806 3.5092 1.3324
024 5.2491 3.9192 1.3333
025 4.3652 2.9279 1.5166
026 5.1675 3.4426 2.0363
027 5.3017 3.4275 1.5963
028 4.1694 2.5320 1.9037
029 4.6223 3.0918 1.6882
030 4.3390 2.7611 1.8345
Averages 6.2160 3.2798 1.7486

Lidar to AGV camera evaluation (different dataset - Dataset_corrected vs Atom_calibration)

rosrun atom_evaluation lidar_to_rgb_evaluation -train_json /home/jorge/datasets/e6_mm/atom_calibration_nig_0.1_0.1.json -test_json /home/jorge/datasets/e6_mm/dataset_corrected.json -cs camera_mb -rs velodyne -si
Collection # RMS (pix) X err (pix) Y err (pix)
000 5.0165 3.1572 1.4601
001 5.5321 3.7744 1.4089
002 5.4894 3.9398 1.4221
003 3.6184 2.5606 1.0372
004 4.6210 3.5103 1.3770
005 2.6547 1.7478 0.7602
006 4.8705 3.3760 1.6965
007 5.8273 3.5149 1.3806
008 4.9740 3.2375 1.3606
009 4.3480 2.8449 1.4139
010 6.3855 4.2268 1.5991
011 4.4527 3.0954 1.4694
012 4.8573 3.4109 1.4984
013 4.1894 2.8002 1.5317
014 4.3140 3.0491 1.4918
015 4.9254 3.1945 1.5025
016 2.1563 1.4797 0.5857
017 14.7812 3.3613 2.8118
018 22.8064 4.7569 5.0110
019 13.5416 4.4927 3.0352
020 4.2044 2.8963 1.3374
021 5.0278 3.3680 1.6414
022 14.8617 4.3641 4.0795
023 4.6806 3.5092 1.3324
024 5.2491 3.9192 1.3333
025 4.3652 2.9279 1.5166
026 5.1675 3.4426 2.0363
027 5.3017 3.4275 1.5963
028 4.1694 2.5320 1.9037
029 4.6223 3.0918 1.6882
030 4.3390 2.7611 1.8345
Averages 6.1726 3.2829 1.7469

Eye-on-hand camera to AGV camera evaluation using different datasets (Dataset_corrected vs Atom_calibration)

rosrun atom_evaluation rgb_to_rgb_evaluation -train_json ~/datasets/e6_mm/atom_calibration_nig_0.1_0.1.json -test_json ~/datasets/e6_mm/dataset_corrected.json -ss camera -st camera_mb -si
Collection # RMS (pix) X err (pix) Y err (pix) Trans (mm) Rot (deg)
000 11.6271 3.1841 11.1411 21.7989 0.4879
001 11.2983 3.0497 10.8327 21.3140 0.6549
002 11.1305 3.2751 10.5919 21.2879 0.5673
003 10.9073 3.0797 10.4086 20.2596 0.5703
004 10.5472 2.7682 10.1234 19.7930 0.5297
005 10.8627 2.9775 10.3927 20.4319 0.2385
006 17.1305 6.7124 14.2736 44.0573 0.2735
007 11.4810 3.2152 10.9508 20.6005 0.2654
008 11.5334 3.2563 11.0081 20.6385 0.2318
009 12.5853 3.7899 11.9621 24.2876 0.3319
010 12.2648 3.6596 11.6532 22.9485 0.3181
011 11.3285 3.1671 10.8179 20.4061 0.3992
012 11.7637 3.4383 11.1925 21.9481 0.2730
013 8.8461 2.8193 8.3511 21.5170 0.2380
014 8.7551 2.8603 8.2447 21.2991 0.2855
015 10.9730 3.0474 10.4928 21.9112 0.2779
016 8.8780 1.9557 8.6431 22.0899 0.2633
017 10.2374 2.3822 9.9146 21.5578 0.1848
018 9.5924 2.4406 9.2279 21.7952 0.2006
019 8.4827 2.5767 8.0577 20.9860 0.3444
020 8.7680 2.7013 8.3145 21.3158 0.2217
021 8.8530 2.7921 8.3737 21.2475 0.2372
022 8.8320 2.7925 8.3514 21.3483 0.1548
023 10.8246 2.1211 10.5724 21.4396 0.1816
024 10.8400 2.7784 10.4344 21.5516 0.1654
025 7.6109 2.8939 7.0050 20.0794 0.3764
026 7.6934 2.6334 7.2030 19.6203 0.1774
027 7.5142 2.5460 7.0364 19.7528 0.3131
028 7.5469 3.2641 6.7746 20.8182 0.9315
029 7.4710 2.7430 6.9286 20.2663 0.7569
030 7.6381 2.5936 7.1571 20.3221 0.5538
Averages 10.1231 3.0166 9.5623 21.8932 0.3550
JorgeFernandes-Git commented 1 year ago

Hi @miguelriemoliveira.

I add a velodyne to the AGV and run some tests. Check it out.

miguelriemoliveira commented 1 year ago

Looks great! But the rms I think also doesmnot make sense when compared to xpix ypix