Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

请问为啥mac上euroc mav数据集会运行一会儿后LM优化挂掉了？ #3

Open
BigDeviltjj opened this issue Nov 15, 2019 · 3 comments
Open

请问为啥mac上euroc mav数据集会运行一会儿后LM优化挂掉了？ #3

BigDeviltjj opened this issue Nov 15, 2019 · 3 comments

Comments

@BigDeviltjj
Copy link

mac系统，数据集是MH_05_difficult, 会运行很短的时间后优化器挂掉，lm算法的lambda值变得很大梯度无法下降。对代码的改动只有把opencv2的宏定义换成opencv4的宏定义，并且检查过应该不会造成影响，关键点一开始也能监测到，请问这个可能的原因是啥？

部分log 如下

/Users/mayday/VINS-Course/bin/run_euroc /Users/mayday/Downloads/mav0/ ../config/
1 System() sConfig_file: ../config/euroc_config.yaml
 fix extrinsic param 
1 readParameters:  
  INIT_DEPTH: 5
  MIN_PARALLAX: 0.0217391
  ACC_N: 0.08
  ACC_W: 4e-05
  GYR_N: 0.004
  GYR_W: 2e-06
  RIC:    0.0148655  -0.999881  0.0041403
  0.999557  0.0149672  0.0257155
-0.0257744 0.00375619   0.999661
  TIC:   -0.0216401  -0.064677 0.00981073
  G:           0       0 9.81007
  BIAS_ACC_THRESHOLD:0.1
  BIAS_GYR_THRESHOLD:0.1
  SOLVER_TIME:0.04
  NUM_ITERATIONS:8
  ESTIMATE_EXTRINSIC:0
  ESTIMATE_TD:0
  ROLLING_SHUTTER:0
  ROW:480
  COL:752
  TD:0
  TR:0
  FOCAL_LENGTH:460
  IMAGE_TOPIC:/cam0/image_raw
  IMU_TOPIC:/imu0
  FISHEYE_MASK:
  CAM_NAMES[0]:../config/euroc_config.yaml
  MAX_CNT:150
  MIN_DIST:30
  FREQ:10
  F_THRESHOLD:1
  SHOW_TRACK:1
  STEREO_TRACK:0
  EQUALIZE:1
  FISHEYE:0
  PUB_THIS_FRAME:0
reading paramerter of camera ../config/euroc_config.yaml
1 PinholeCamera 
  m_cameraName: camera
  m_imageWidth: 752
  m_imageHeight: 480
  m_k1: -0.2917
  m_k2: 0.08228
  m_p1: 5.333e-05
  m_p2: -0.0001578
  m_fx: 461.6
  m_fy: 460.3
  m_cx: 363
  m_cy: 248.1

1 Estimator::setParameter FOCAL_LENGTH: 460
2 System() end
1 ProcessBackEnd start
1 PubImuData start sImu_data_filea: ../config/MH_05_imu0.txt1 PubImageData start sImage_file: 1 PubImageData start sImage_file: 
../config/MH_05_cam0.txt
../config/MH_05_cam0.txt
1 PubImageData skip the first detected feature, which doesn't contain optical flow speed
2 PubImageData first_image_flag
2019-11-16 00:19:25.191527+0800 run_euroc[5569:2277730] MessageTracer: Falling back to default whitelist
4 PubImage init_pub skip the first image!
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 0
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 1
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 2
wait for imu, only should happen at the beginning sum_of_wait: 3
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 4
wait for imu, only should happen at the beginning sum_of_wait: 5
wait for imu, only should happen at the beginning sum_of_wait: 6
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 7
wait for imu, only should happen at the beginning sum_of_wait: 8
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 9
wait for imu, only should happen at the beginning sum_of_wait: 10
wait for imu, only should happen at the beginning sum_of_wait: 11
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 12
wait for imu, only should happen at the beginning sum_of_wait: 13
wait for imu, only should happen at the beginning sum_of_wait: 14
wait for imu, only should happen at the beginning sum_of_wait: 15
wait for imu, only should happen at the beginning sum_of_wait: 16
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 17
wait for imu, only should happen at the beginning sum_of_wait: 18
wait for imu, only should happen at the beginning sum_of_wait: 19
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 20
wait for imu, only should happen at the beginning sum_of_wait: 21
wait for imu, only should happen at the beginning sum_of_wait: 22
wait for imu, only should happen at the beginning sum_of_wait: 23
wait for imu, only should happen at the beginning sum_of_wait: 24
wait for imu, only should happen at the beginning sum_of_wait: 25
wait for imu, only should happen at the beginning sum_of_wait: 26
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 27
wait for imu, only should happen at the beginning sum_of_wait: 28
wait for imu, only should happen at the beginning sum_of_wait: 29
wait for imu, only should happen at the beginning sum_of_wait: 30
wait for imu, only should happen at the beginning sum_of_wait: 31
wait for imu, only should happen at the beginning sum_of_wait: 32
wait for imu, only should happen at the beginning sum_of_wait: 33
wait for imu, only should happen at the beginning sum_of_wait: 34
wait for imu, only should happen at the beginning sum_of_wait: 35
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 36
wait for imu, only should happen at the beginning sum_of_wait: 37
wait for imu, only should happen at the beginning sum_of_wait: 38
wait for imu, only should happen at the beginning sum_of_wait: 39
wait for imu, only should happen at the beginning sum_of_wait: 40
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 41
wait for imu, only should happen at the beginning sum_of_wait: 42
wait for imu, only should happen at the beginning sum_of_wait: 43
wait for imu, only should happen at the beginning sum_of_wait: 44
wait for imu, only should happen at the beginning sum_of_wait: 45
wait for imu, only should happen at the beginning sum_of_wait: 46
wait for imu, only should happen at the beginning sum_of_wait: 47
wait for imu, only should happen at the beginning sum_of_wait: 48
wait for imu, only should happen at the beginning sum_of_wait: 49
wait for imu, only should happen at the beginning sum_of_wait: 50
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 51
wait for imu, only should happen at the beginning sum_of_wait: 52
wait for imu, only should happen at the beginning sum_of_wait: 53
wait for imu, only should happen at the beginning sum_of_wait: 54
wait for imu, only should happen at the beginning sum_of_wait: 55
wait for imu, only should happen at the beginning sum_of_wait: 56
wait for imu, only should happen at the beginning sum_of_wait: 57
wait for imu, only should happen at the beginning sum_of_wait: 58
wait for imu, only should happen at the beginning sum_of_wait: 59
wait for imu, only should happen at the beginning sum_of_wait: 60
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 61
wait for imu, only should happen at the beginning sum_of_wait: 62
wait for imu, only should happen at the beginning sum_of_wait: 63
wait for imu, only should happen at the beginning sum_of_wait: 64
wait for imu, only should happen at the beginning sum_of_wait: 65
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 66
wait for imu, only should happen at the beginning sum_of_wait: 67
wait for imu, only should happen at the beginning sum_of_wait: 68
wait for imu, only should happen at the beginning sum_of_wait: 69
wait for imu, only should happen at the beginning sum_of_wait: 70
wait for imu, only should happen at the beginning sum_of_wait: 71
wait for imu, only should happen at the beginning sum_of_wait: 72
wait for imu, only should happen at the beginning sum_of_wait: 73
wait for imu, only should happen at the beginning sum_of_wait: 74
wait for imu, only should happen at the beginning sum_of_wait: 75
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 76
wait for imu, only should happen at the beginning sum_of_wait: 77
wait for imu, only should happen at the beginning sum_of_wait: 78
wait for imu, only should happen at the beginning sum_of_wait: 79
wait for imu, only should happen at the beginning sum_of_wait: 80
wait for imu, only should happen at the beginning sum_of_wait: 81
wait for imu, only should happen at the beginning sum_of_wait: 82
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 83
wait for imu, only should happen at the beginning sum_of_wait: 84
wait for imu, only should happen at the beginning sum_of_wait: 85
Not enough features or parallax; Move device around
wait for imu, only should happen at the beginning sum_of_wait: 86
wait for imu, only should happen at the beginning sum_of_wait: 87
wait for imu, only should happen at the beginning sum_of_wait: 88
wait for imu, only should happen at the beginning sum_of_wait: 89
global SFM failed!
wait for imu, only should happen at the beginning sum_of_wait: 90
global SFM failed!
global SFM failed!
global SFM failed!
iter: 0 , chi= 3.21312e+06 , Lambda= 500000
iter: 1 , chi= 8210.04 , Lambda= 333333
iter: 2 , chi= 4502.87 , Lambda= 222222
iter: 3 , chi= 3173.47 , Lambda= 148148
iter: 4 , chi= 2345.79 , Lambda= 98765.4
iter: 5 , chi= 1740.97 , Lambda= 65843.6
iter: 6 , chi= 1275.09 , Lambda= 43895.7
iter: 7 , chi= 970.319 , Lambda= 29263.8
iter: 8 , chi= 800.372 , Lambda= 19509.2
iter: 9 , chi= 725.164 , Lambda= 13006.1
problem solve cost: 204.265 ms
   makeHessian cost: 150.437 ms
edge factor cnt: 578
Initialization finish!
1 BackEnd processImage dt: 318.181000 stamp: 1403638523.027829 p_wi:  0.016292 -0.002303  0.130714
iter: 0 , chi= 745.454920 , Lambda= 500000.000000
iter: 1 , chi= 692.065917 , Lambda= 333333.333333
iter: 2 , chi= 661.554274 , Lambda= 222222.222222
iter: 3 , chi= 644.234796 , Lambda= 148148.148148
iter: 4 , chi= 630.264263 , Lambda= 98765.432099
iter: 5 , chi= 616.299235 , Lambda= 65843.621399
iter: 6 , chi= 600.444146 , Lambda= 43895.747599
iter: 7 , chi= 581.436377 , Lambda= 29263.831733
iter: 8 , chi= 558.663966 , Lambda= 19509.221155
iter: 9 , chi= 532.128932 , Lambda= 13006.147437
problem solve cost: 187.251000 ms
   makeHessian cost: 146.118000 ms
----------- update bprior -------------
             before: 17386.725580
                     10.671542
             after: 14626.155381
                    7.425029
edge factor cnt: 0
1 BackEnd processImage dt: 206.308000 stamp: 1403638523.127830 p_wi:  0.017609 -0.006107  0.168797
iter: 0 , chi= 592.010783 , Lambda= 500000.000000
iter: 1 , chi= 582.011298 , Lambda= 333333.333333
iter: 2 , chi= 578.244231 , Lambda= 222222.222222
iter: 3 , chi= 573.568290 , Lambda= 148148.148148
iter: 4 , chi= 569.382454 , Lambda= 98765.432099
iter: 5 , chi= 564.178063 , Lambda= 65843.621399
iter: 6 , chi= 557.947238 , Lambda= 43895.747599
iter: 7 , chi= 551.057875 , Lambda= 29263.831733
iter: 8 , chi= 544.329040 , Lambda= 19509.221155
iter: 9 , chi= 538.798409 , Lambda= 13006.147437
problem solve cost: 183.791000 ms
   makeHessian cost: 142.183000 ms
----------- update bprior -------------
             before: 14169.987091
                     7.239563
             after: 14003.869986
                    6.498913
edge factor cnt: 470
1 BackEnd processImage dt: 221.172000 stamp: 1403638523.227829 p_wi:  0.014177 -0.012528  0.165095
iter: 0 , chi= 694.372067 , Lambda= 500000.000000
iter: 1 , chi= 601.031408 , Lambda= 333333.333333
iter: 2 , chi= 594.941343 , Lambda= 222222.222222
iter: 3 , chi= 583.987674 , Lambda= 148148.148148
iter: 4 , chi= 576.888224 , Lambda= 98765.432099
iter: 5 , chi= 569.774742 , Lambda= 65843.621399
iter: 6 , chi= 565.930514 , Lambda= 43895.747599
iter: 7 , chi= 562.297411 , Lambda= 29263.831733
iter: 8 , chi= 559.073784 , Lambda= 19509.221155
iter: 9 , chi= 556.178606 , Lambda= 13006.147437
problem solve cost: 213.803000 ms
   makeHessian cost: 161.372000 ms
----------- update bprior -------------
             before: 42414.732696
                     13.576271
             after: 50182.140686
                    13.855544
edge factor cnt: 0
1 BackEnd processImage dt: 232.737000 stamp: 1403638523.327830 p_wi:  0.009445 -0.019707  0.129923
iter: 0 , chi= 618.968552 , Lambda= 500000.000000
iter: 1 , chi= 613.714480 , Lambda= 333333.333333
iter: 2 , chi= 612.751168 , Lambda= 222222.222222
iter: 3 , chi= 611.939101 , Lambda= 148148.148148
iter: 4 , chi= 611.117423 , Lambda= 98765.432099
iter: 5 , chi= 610.941737 , Lambda= 65843.621399
iter: 6 , chi= 610.798154 , Lambda= 43895.747599
iter: 7 , chi= 610.644269 , Lambda= 29263.831733
iter: 8 , chi= 610.486812 , Lambda= 19509.221155
iter: 9 , chi= 610.338178 , Lambda= 13006.147437
problem solve cost: 203.063000 ms
   makeHessian cost: 157.687000 ms
----------- update bprior -------------
             before: 50340.959920
                     13.267921
             after: 51582.647278
                    13.446301
edge factor cnt: 448
1 BackEnd processImage dt: 236.664000 stamp: 1403638523.427830 p_wi:  0.004868 -0.027164  0.066776
iter: 0 , chi= 842.197330 , Lambda= 500000.000000
iter: 1 , chi= 783.428681 , Lambda= 333333.333333
iter: 2 , chi= 780.847695 , Lambda= 222222.222222
iter: 3 , chi= 779.764199 , Lambda= 148148.148148
iter: 4 , chi= 777.817808 , Lambda= 98765.432099
iter: 5 , chi= 776.614725 , Lambda= 65843.621399
iter: 6 , chi= 775.221496 , Lambda= 43895.747599
iter: 7 , chi= 773.612238 , Lambda= 29263.831733
iter: 8 , chi= 771.890227 , Lambda= 19509.221155
iter: 9 , chi= 770.232471 , Lambda= 13006.147437
problem solve cost: 204.635000 ms
   makeHessian cost: 160.102000 ms
----------- update bprior -------------
             before: 32990.497548
                     16.372910
             after: 36348.958273
                    16.900114
edge factor cnt: 0
1 BackEnd processImage dt: 221.618000 stamp: 1403638523.527829 p_wi: -0.002942 -0.034283  0.004429
iter: 0 , chi= 801.218070 , Lambda= 500000.000000
iter: 1 , chi= 798.604132 , Lambda= 333333.333333
iter: 2 , chi= 797.858588 , Lambda= 222222.222222
iter: 3 , chi= 796.842739 , Lambda= 148148.148148
iter: 4 , chi= 796.007481 , Lambda= 98765.432099
iter: 5 , chi= 795.168622 , Lambda= 65843.621399
iter: 6 , chi= 794.910050 , Lambda= 43895.747599
iter: 7 , chi= 794.839043 , Lambda= 29263.831733
iter: 8 , chi= 794.757570 , Lambda= 19509.221155
iter: 9 , chi= 794.655636 , Lambda= 13006.147437
problem solve cost: 194.144000 ms
   makeHessian cost: 151.773000 ms
----------- update bprior -------------
             before: 36218.304054
                     16.132964
             after: 35851.569961
                    16.247883
edge factor cnt: 467
1 BackEnd processImage dt: 230.390000 stamp: 1403638523.627830 p_wi: -0.013165 -0.042818 -0.072382
iter: 0 , chi= 1069.533261 , Lambda= 500000.000000
iter: 1 , chi= 1053.418973 , Lambda= 333333.333333
iter: 2 , chi= 1048.820284 , Lambda= 222222.222222
iter: 3 , chi= 1045.711340 , Lambda= 148148.148148
iter: 4 , chi= 1043.434903 , Lambda= 98765.432099
iter: 5 , chi= 1039.571154 , Lambda= 65843.621399
iter: 6 , chi= 1037.512597 , Lambda= 43895.747599
iter: 7 , chi= 1035.312532 , Lambda= 29263.831733
iter: 8 , chi= 1032.699930 , Lambda= 19509.221155
iter: 9 , chi= 1029.609448 , Lambda= 13006.147437
problem solve cost: 196.057000 ms
   makeHessian cost: 155.347000 ms
----------- update bprior -------------
             before: 33413.110198
                     20.197264
             after: 39000.148790
                    20.790668
edge factor cnt: 591
1 BackEnd processImage dt: 246.134000 stamp: 1403638523.727829 p_wi: -0.025514 -0.049371 -0.132863
iter: 0 , chi= 1310.138242 , Lambda= 500000.000000
iter: 1 , chi= 1290.254593 , Lambda= 333333.333333
iter: 2 , chi= 1286.818376 , Lambda= 222222.222222
iter: 3 , chi= 1283.928965 , Lambda= 148148.148148
iter: 4 , chi= 1280.561228 , Lambda= 98765.432099
iter: 5 , chi= 1276.329520 , Lambda= 65843.621399
iter: 6 , chi= 1270.692393 , Lambda= 43895.747599
iter: 7 , chi= 1262.817790 , Lambda= 29263.831733
iter: 8 , chi= 1254.598696 , Lambda= 19509.221155
iter: 9 , chi= 1246.756114 , Lambda= 13006.147437
problem solve cost: 198.699000 ms
   makeHessian cost: 160.930000 ms
----------- update bprior -------------
             before: 46838.490385
                     25.318841
             after: 54468.008983
                    25.717354
edge factor cnt: 730
1 BackEnd processImage dt: 241.043000 stamp: 1403638523.827830 p_wi: -0.041008 -0.054122 -0.179356
iter: 0 , chi= 1550.392285 , Lambda= 500000.000000
iter: 1 , chi= 1502.995983 , Lambda= 333333.333333
iter: 2 , chi= 1484.334354 , Lambda= 222222.222222
iter: 3 , chi= 1467.606615 , Lambda= 148148.148148
iter: 4 , chi= 1451.448693 , Lambda= 98765.432099
iter: 5 , chi= 1434.425435 , Lambda= 65843.621399
iter: 6 , chi= 1415.721837 , Lambda= 43895.747599
iter: 7 , chi= 1399.357186 , Lambda= 29263.831733
iter: 8 , chi= 1384.925155 , Lambda= 19509.221155
iter: 9 , chi= 1375.210939 , Lambda= 13006.147437
problem solve cost: 187.321000 ms
   makeHessian cost: 148.856000 ms
----------- update bprior -------------
             before: 59873.503809
                     31.473237
             after: 68567.282949
                    31.286783
edge factor cnt: 790
1 BackEnd processImage dt: 228.961000 stamp: 1403638523.927830 p_wi: -0.058104 -0.062289 -0.214709
iter: 0 , chi= 1603.453001 , Lambda= 500000.000000
iter: 1 , chi= 1525.481130 , Lambda= 333333.333333
iter: 2 , chi= 1470.810206 , Lambda= 222222.222222
iter: 3 , chi= 1405.662624 , Lambda= 148148.148148
iter: 4 , chi= 1363.673877 , Lambda= 98765.432099
iter: 5 , chi= 1343.250289 , Lambda= 65843.621399
iter: 6 , chi= 1332.327246 , Lambda= 43895.747599
iter: 7 , chi= 1326.844510 , Lambda= 29263.831733
iter: 8 , chi= 1322.182442 , Lambda= 19509.221155
iter: 9 , chi= 1316.557399 , Lambda= 13006.147437
problem solve cost: 216.895000 ms
   makeHessian cost: 172.423000 ms
----------- update bprior -------------
             before: 87490.668901
                     37.334423
             after: 80645.319205
                    36.450623
edge factor cnt: 838
1 BackEnd processImage dt: 269.576000 stamp: 1403638524.027829 p_wi: -0.067759 -0.069016 -0.244528
iter: 0 , chi= 1434.323578 , Lambda= 500000.000000
iter: 1 , chi= 1391.105990 , Lambda= 333333.333333
iter: 2 , chi= 1377.726917 , Lambda= 222222.222222
iter: 3 , chi= 1367.367963 , Lambda= 148148.148148
iter: 4 , chi= 1358.269425 , Lambda= 98765.432099
iter: 5 , chi= 1350.909867 , Lambda= 65843.621399
iter: 6 , chi= 1347.816204 , Lambda= 43895.747599
iter: 7 , chi= 1345.395316 , Lambda= 29263.831733
iter: 8 , chi= 1342.559157 , Lambda= 19509.221155
iter: 9 , chi= 1339.455141 , Lambda= 13006.147437
problem solve cost: 196.125000 ms
   makeHessian cost: 154.938000 ms
----------- update bprior -------------
             before: 114059.941258
                     40.964931
             after: 99023.065193
                    40.194007
edge factor cnt: 837
1 BackEnd processImage dt: 242.571000 stamp: 1403638524.127830 p_wi: -0.069669 -0.075533 -0.254451
iter: 0 , chi= 1536.097765 , Lambda= 500000.000000
iter: 1 , chi= 1494.664579 , Lambda= 333333.333333
iter: 2 , chi= 1490.029223 , Lambda= 222222.222222
iter: 3 , chi= 1484.799843 , Lambda= 148148.148148
iter: 4 , chi= 1482.651273 , Lambda= 98765.432099
iter: 5 , chi= 1480.168687 , Lambda= 65843.621399
iter: 6 , chi= 1479.650408 , Lambda= 43895.747599
iter: 7 , chi= 1475.177242 , Lambda= 29263.831733
iter: 8 , chi= 1471.207998 , Lambda= 19509.221155
iter: 9 , chi= 1468.554312 , Lambda= 13006.147437
problem solve cost: 200.785000 ms
   makeHessian cost: 157.558000 ms
----------- update bprior -------------
             before: 127069.810658
                     43.272132
             after: 115763.106860
                    42.922662
edge factor cnt: 790
1 BackEnd processImage dt: 247.849000 stamp: 1403638524.227829 p_wi: -0.068546 -0.080281 -0.236345
iter: 0 , chi= 1709.258689 , Lambda= 500000.000000
iter: 1 , chi= 1615.207860 , Lambda= 333333.333333
iter: 2 , chi= 1609.037681 , Lambda= 222222.222222
iter: 3 , chi= 1603.147548 , Lambda= 148148.148148
iter: 4 , chi= 1601.114388 , Lambda= 98765.432099
iter: 5 , chi= 1600.313477 , Lambda= 65843.621399
iter: 6 , chi= 1600.199016 , Lambda= 43895.747599
iter: 7 , chi= 1597.650772 , Lambda= 29263.831733
iter: 8 , chi= 1595.044048 , Lambda= 19509.221155
iter: 9 , chi= 1592.498781 , Lambda= 13006.147437
problem solve cost: 215.490000 ms
   makeHessian cost: 166.501000 ms
----------- update bprior -------------
             before: 113781.819664
                     44.155281
             after: 115444.634657
                    44.264920
edge factor cnt: 753
1 BackEnd processImage dt: 257.131000 stamp: 1403638524.327830 p_wi: -0.067473 -0.080223 -0.197262
iter: 0 , chi= 1440.931473 , Lambda= 500000.000000
iter: 1 , chi= 1425.241460 , Lambda= 333333.333333
iter: 2 , chi= 1422.119414 , Lambda= 222222.222222
iter: 3 , chi= 1420.777400 , Lambda= 148148.148148
iter: 4 , chi= 1419.684935 , Lambda= 98765.432099
iter: 5 , chi= 1418.347581 , Lambda= 65843.621399
iter: 6 , chi= 1416.455510 , Lambda= 43895.747599
iter: 7 , chi= 1413.859472 , Lambda= 29263.831733
iter: 8 , chi= 1410.346409 , Lambda= 19509.221155
sqrt(currentChi_) <= stopThresholdLM_
problem solve cost: 196.886000 ms
   makeHessian cost: 127.699000 ms
----------- update bprior -------------
             before: 69770.568497
                     38.881166
             after: 72689.998650
                    38.788844
edge factor cnt: 689
1 BackEnd processImage dt: 237.178000 stamp: 1403638524.427830 p_wi: -0.067647 -0.079907 -0.152648
iter: 0 , chi= 1231.593173 , Lambda= 500000.000000
iter: 1 , chi= 1231.593173 , Lambda= 18014398509481986097152.000000
sqrt(currentChi_) <= stopThresholdLM_
problem solve cost: 83.356000 ms
   makeHessian cost: 15.171000 ms
----------- update bprior -------------
             before: 46436.255845
                     31.919516
             after: 46436.255845
                    31.919516
edge factor cnt: 641
1 BackEnd processImage dt: 119.542000 stamp: 1403638524.527829 p_wi: -0.070737 -0.080112 -0.112700
iter: 0 , chi= 1288.517295 , Lambda= 500000.000000
iter: 1 , chi= 1288.517295 , Lambda= 18014398509481986097152.000000
sqrt(currentChi_) <= stopThresholdLM_
problem solve cost: 73.814000 ms
   makeHessian cost: 12.162000 ms
----------- update bprior -------------
             before: 49913.403486
                     31.370271
             after: 49913.403486
                    31.370271
edge factor cnt: 650
1 BackEnd processImage dt: 111.780000 stamp: 1403638524.627830 p_wi: -0.073386 -0.078920 -0.076434
iter: 0 , chi= 1422.796583 , Lambda= 500000.000000
iter: 1 , chi= 1422.796583 , Lambda= 18014398509481986097152.000000
sqrt(currentChi_) <= stopThresholdLM_
problem solve cost: 81.417000 ms
   makeHessian cost: 12.464000 ms
----------- update bprior -------------
             before: 63233.286556
                     32.801510
             after: 63233.286556
                    32.801510
edge factor cnt: 621
1 BackEnd processImage dt: 121.892000 stamp: 1403638524.727829 p_wi: -0.077202 -0.075908 -0.050835
iter: 0 , chi= 1621.623890 , Lambda= 500000.000000
iter: 1 , chi= 1621.623890 , Lambda= 18014398509481986097152.000000
sqrt(currentChi_) <= stopThresholdLM_
problem solve cost: 91.428000 ms
   makeHessian cost: 16.504000 ms
----------- update bprior -------------
             before: 79747.458278
                     35.574162
             after: 79747.458278
                    35.574162
edge factor cnt: 604
1 BackEnd processImage dt: 130.321000 stamp: 1403638524.827830 p_wi: -0.083894 -0.072473 -0.040443
iter: 0 , chi= 1875.906090 , Lambda= 500000.000000
iter: 1 , chi= 1875.906090 , Lambda= 18014398509481986097152.000000
sqrt(currentChi_) <= stopThresholdLM_
problem solve cost: 91.879000 ms
   makeHessian cost: 13.470000 ms
----------- update bprior -------------
             before: 116926.465427
                     39.161414
             after: 116926.465427
                    39.161414
edge factor cnt: 637
1 BackEnd processImage dt: 127.697000 stamp: 1403638524.927830 p_wi: -0.092988 -0.068653 -0.045473
iter: 0 , chi= 2187.102491 , Lambda= 500000.000000
iter: 1 , chi= 2187.102491 , Lambda= 18014398509481986097152.000000
sqrt(currentChi_) <= stopThresholdLM_
problem solve cost: 99.919000 ms
   makeHessian cost: 14.222000 ms
----------- update bprior -------------
             before: 174055.355823
                     43.478496
             after: 174055.355823
                    43.478496
edge factor cnt: 0
1 BackEnd processImage dt: 117.768000 stamp: 1403638525.027829 p_wi: -0.106233 -0.066695 -0.070588
iter: 0 , chi= 2228.768144 , Lambda= 500000.000000
iter: 1 , chi= 2228.768144 , Lambda= 18014398509481986097152.000000
sqrt(currentChi_) <= stopThresholdLM_
problem solve cost: 96.634000 ms
   makeHessian cost: 13.822000 ms
----------- update bprior -------------
             before: 171588.900937
                     43.034115
             after: 171588.900937
                    43.034115
edge factor cnt: 648
1 BackEnd processImage dt: 135.358000 stamp: 1403638525.127830 p_wi: -0.120505 -0.065431 -0.119497
iter: 0 , chi= 2377.880994 , Lambda= 500000.000000
iter: 1 , chi= 2377.880994 , Lambda= 18014398509481986097152.000000
sqrt(currentChi_) <= stopThresholdLM_
problem solve cost: 97.825000 ms
   makeHessian cost: 14.555000 ms
----------- update bprior -------------
             before: 168171.583049
                     44.160296
             after: 168171.583049
                    44.160296
edge factor cnt: 0
1 BackEnd processImage dt: 115.886000 stamp: 1403638525.227829 p_wi: -0.136085 -0.064623 -0.186639
iter: 0 , chi= 2395.283405 , Lambda= 500000.000000
iter: 1 , chi= 2395.283405 , Lambda= 18014398509481986097152.000000
sqrt(currentChi_) <= stopThresholdLM_
problem solve cost: 107.141000 ms
   makeHessian cost: 15.033000 ms
----------- update bprior -------------
             before: 167781.619606
                     43.904490
             after: 167781.619606
                    43.904490
edge factor cnt: 0
1 BackEnd processImage dt: 123.770000 stamp: 1403638525.327830 p_wi: -0.154783 -0.063897 -0.264847
iter: 0 , chi= 2394.974157 , Lambda= 500000.000000
iter: 1 , chi= 2394.974157 , Lambda= 18014398509481986097152.000000
sqrt(currentChi_) <= stopThresholdLM_
problem solve cost: 98.519000 ms
   makeHessian cost: 13.840000 ms
----------- update bprior -------------
             before: 167781.619606
                     43.904490
             after: 167781.619606
                    43.904490
edge factor cnt: 594
1 BackEnd processImage dt: 232.593000 stamp: 1403638525.427830 p_wi: -0.176794 -0.061802 -0.346876
iter: 0 , chi= nan , Lambda= 500000.000000
iter: 1 , chi= nan , Lambda= 18014398509481986097152.000000
iter: 2 , chi= nan , Lambda= 822752278660603116858455895396728899888057129711113907845499648475136.000000
iter: 3 , chi= nan , Lambda= 47634102635436898724379873315377984320228608500171483271481892445135526726758807532347872310828455112538944824267416431834545575140033753564315648.000000
iter: 4 , chi= nan , Lambda= 3495959950985713444632080123730709903643510074804632702858961286653459985776588139814929586748127946306818840941912242619549823910362988237896222302398330053978381405443632220440135622373810791965842646989594435053520339544682443171476172588464958603264.000000
iter: 5 , chi= nan , Lambda= inf
iter: 6 , chi= nan , Lambda= inf
iter: 7 , chi= nan , Lambda= inf
iter: 8 , chi= nan , Lambda= inf
iter: 9 , chi= nan , Lambda= inf
problem solve cost: 405.440000 ms
   makeHessian cost: 13.611000 ms
----------- update bprior -------------
             before: nan
                     nan
             after: nan
                    nan
edge factor cnt: 530
1 BackEnd processImage dt: 440.160000 stamp: 1403638525.527829 p_wi: -0.201462 -0.059313 -0.426397
iter: 0 , chi= nan , Lambda= 500000.000000
iter: 1 , chi= nan , Lambda= 18014398509481986097152.000000
iter: 2 , chi= nan , Lambda= 822752278660603116858455895396728899888057129711113907845499648475136.000000
iter: 3 , chi= nan , Lambda= 47634102635436898724379873315377984320228608500171483271481892445135526726758807532347872310828455112538944824267416431834545575140033753564315648.000000
iter: 4 , chi= nan , Lambda= 3495959950985713444632080123730709903643510074804632702858961286653459985776588139814929586748127946306818840941912242619549823910362988237896222302398330053978381405443632220440135622373810791965842646989594435053520339544682443171476172588464958603264.000000
iter: 5 , chi= nan , Lambda= inf
iter: 6 , chi= nan , Lambda= inf
iter: 7 , chi= nan , Lambda= inf
iter: 8 , chi= nan , Lambda= inf
iter: 9 , chi= nan , Lambda= inf
problem solve cost: 361.049000 ms
   makeHessian cost: 16.291000 ms
----------- update bprior -------------
             before: nan
                     nan
             after: nan
                    nan
edge factor cnt: 561
1 BackEnd processImage dt: 396.373000 stamp: 1403638525.627830 p_wi: -0.226856 -0.058145 -0.499777
1 getMeasurements size: 2 imu sizes: 21 feature_buf size: 0 imu_buf size: 48
iter: 0 , chi= nan , Lambda= 500000.000000
iter: 1 , chi= nan , Lambda= 18014398509481986097152.000000
iter: 2 , chi= nan , Lambda= 822752278660603116858455895396728899888057129711113907845499648475136.000000
iter: 3 , chi= nan , Lambda= 47634102635436898724379873315377984320228608500171483271481892445135526726758807532347872310828455112538944824267416431834545575140033753564315648.000000
iter: 4 , chi= nan , Lambda= 3495959950985713444632080123730709903643510074804632702858961286653459985776588139814929586748127946306818840941912242619549823910362988237896222302398330053978381405443632220440135622373810791965842646989594435053520339544682443171476172588464958603264.000000
iter: 5 , chi= nan , Lambda= inf
iter: 6 , chi= nan , Lambda= inf
iter: 7 , chi= nan , Lambda= inf
iter: 8 , chi= nan , Lambda= inf
iter: 9 , chi= nan , Lambda= inf
problem solve cost: 378.360000 ms
   makeHessian cost: 16.244000 ms
----------- update bprior -------------
             before: nan
                     nan
             after: nan
                    nan
edge factor cnt: 722
1 BackEnd processImage dt: 418.112000 stamp: 1403638525.727829 p_wi: -0.251868 -0.057143 -0.565037
iter: 0 , chi= nan , Lambda= 500000.000000
iter: 1 , chi= nan , Lambda= 18014398509481986097152.000000
iter: 2 , chi= nan , Lambda= 822752278660603116858455895396728899888057129711113907845499648475136.000000
iter: 3 , chi= nan , Lambda= 47634102635436898724379873315377984320228608500171483271481892445135526726758807532347872310828455112538944824267416431834545575140033753564315648.000000
iter: 4 , chi= nan , Lambda= 3495959950985713444632080123730709903643510074804632702858961286653459985776588139814929586748127946306818840941912242619549823910362988237896222302398330053978381405443632220440135622373810791965842646989594435053520339544682443171476172588464958603264.000000
iter: 5 , chi= nan , Lambda= inf
iter: 6 , chi= nan , Lambda= inf
iter: 7 , chi= nan , Lambda= inf
iter: 8 , chi= nan , Lambda= inf
iter: 9 , chi= nan , Lambda= inf
problem solve cost: 368.905000 ms
   makeHessian cost: 14.196000 ms
----------- update bprior -------------
             before: nan
                     nan
             after: nan
                    nan
edge factor cnt: 756
1 BackEnd processImage dt: 408.779000 stamp: 1403638525.827830 p_wi: -0.277331 -0.055067 -0.621944
1 getMeasurements size: 4 imu sizes: 21 feature_buf size: 0 imu_buf size: 43
iter: 0 , chi= nan , Lambda= 500000.000000
iter: 1 , chi= nan , Lambda= 18014398509481986097152.000000
iter: 2 , chi= nan , Lambda= 822752278660603116858455895396728899888057129711113907845499648475136.000000
iter: 3 , chi= nan , Lambda= 47634102635436898724379873315377984320228608500171483271481892445135526726758807532347872310828455112538944824267416431834545575140033753564315648.000000
iter: 4 , chi= nan , Lambda= 3495959950985713444632080123730709903643510074804632702858961286653459985776588139814929586748127946306818840941912242619549823910362988237896222302398330053978381405443632220440135622373810791965842646989594435053520339544682443171476172588464958603264.000000
iter: 5 , chi= nan , Lambda= inf
iter: 6 , chi= nan , Lambda= inf
iter: 7 , chi= nan , Lambda= inf
iter: 8 , chi= nan , Lambda= inf
iter: 9 , chi= nan , Lambda= inf
problem solve cost: 386.377000 ms
   makeHessian cost: 14.996000 ms
----------- update bprior -------------
             before: nan
                     nan
             after: nan
                    nan
edge factor cnt: 821
1 BackEnd processImage dt: 431.753000 stamp: 1403638525.927830 p_wi: -0.304992 -0.055925 -0.669915
iter: 0 , chi= nan , Lambda= 500000.000000
iter: 1 , chi= nan , Lambda= 18014398509481986097152.000000
iter: 2 , chi= nan , Lambda= 822752278660603116858455895396728899888057129711113907845499648475136.000000
iter: 3 , chi= nan , Lambda= 47634102635436898724379873315377984320228608500171483271481892445135526726758807532347872310828455112538944824267416431834545575140033753564315648.000000
iter: 4 , chi= nan , Lambda= 3495959950985713444632080123730709903643510074804632702858961286653459985776588139814929586748127946306818840941912242619549823910362988237896222302398330053978381405443632220440135622373810791965842646989594435053520339544682443171476172588464958603264.000000
iter: 5 , chi= nan , Lambda= inf
iter: 6 , chi= nan , Lambda= inf
iter: 7 , chi= nan , Lambda= inf
iter: 8 , chi= nan , Lambda= inf
iter: 9 , chi= nan , Lambda= inf
problem solve cost: 420.134000 ms
   makeHessian cost: 13.650000 ms
----------- update bprior -------------
             before: nan
                     nan
             after: nan
                    nan
edge factor cnt: 785
1 BackEnd processImage dt: 461.705000 stamp: 1403638526.027829 p_wi: -0.335642 -0.066689 -0.708128
iter: 0 , chi= nan , Lambda= 500000.000000
iter: 1 , chi= nan , Lambda= 18014398509481986097152.000000
iter: 2 , chi= nan , Lambda= 822752278660603116858455895396728899888057129711113907845499648475136.000000
iter: 3 , chi= nan , Lambda= 47634102635436898724379873315377984320228608500171483271481892445135526726758807532347872310828455112538944824267416431834545575140033753564315648.000000
iter: 4 , chi= nan , Lambda= 3495959950985713444632080123730709903643510074804632702858961286653459985776588139814929586748127946306818840941912242619549823910362988237896222302398330053978381405443632220440135622373810791965842646989594435053520339544682443171476172588464958603264.000000
iter: 5 , chi= nan , Lambda= inf
iter: 6 , chi= nan , Lambda= inf
iter: 7 , chi= nan , Lambda= inf
iter: 8 , chi= nan , Lambda= inf
iter: 9 , chi= nan , Lambda= inf
problem solve cost: 447.874000 ms
   makeHessian cost: 15.610000 ms
----------- update bprior -------------
             before: nan
                     nan
             after: nan
                    nan
edge factor cnt: 746
1 BackEnd processImage dt: 492.594000 stamp: 1403638526.127830 p_wi: -0.367153 -0.088407 -0.738354
iter: 0 , chi= nan , Lambda= 500000.000000
iter: 1 , chi= nan , Lambda= 18014398509481986097152.000000
iter: 2 , chi= nan , Lambda= 822752278660603116858455895396728899888057129711113907845499648475136.000000
iter: 3 , chi= nan , Lambda= 47634102635436898724379873315377984320228608500171483271481892445135526726758807532347872310828455112538944824267416431834545575140033753564315648.000000
iter: 4 , chi= nan , Lambda= 3495959950985713444632080123730709903643510074804632702858961286653459985776588139814929586748127946306818840941912242619549823910362988237896222302398330053978381405443632220440135622373810791965842646989594435053520339544682443171476172588464958603264.000000
iter: 5 , chi= nan , Lambda= inf
iter: 6 , chi= nan , Lambda= inf
iter: 7 , chi= nan , Lambda= inf
iter: 8 , chi= nan , Lambda= inf
iter: 9 , chi= nan , Lambda= inf
problem solve cost: 471.397000 ms
   makeHessian cost: 15.090000 ms
----------- update bprior -------------
             before: nan
                     nan
             after: nan
                    nan
edge factor cnt: 690
1 BackEnd processImage dt: 515.647000 stamp: 1403638526.227829 p_wi: -0.400756 -0.118308 -0.761284
1 getMeasurements size: 8 imu sizes: 21 feature_buf size: 0 imu_buf size: 52
iter: 0 , chi= nan , Lambda= 500000.000000
iter: 1 , chi= nan , Lambda= 18014398509481986097152.000000
iter: 2 , chi= nan , Lambda= 822752278660603116858455895396728899888057129711113907845499648475136.000000
iter: 3 , chi= nan , Lambda= 47634102635436898724379873315377984320228608500171483271481892445135526726758807532347872310828455112538944824267416431834545575140033753564315648.000000
iter: 4 , chi= nan , Lambda= 3495959950985713444632080123730709903643510074804632702858961286653459985776588139814929586748127946306818840941912242619549823910362988237896222302398330053978381405443632220440135622373810791965842646989594435053520339544682443171476172588464958603264.000000
iter: 5 , chi= nan , Lambda= inf
iter: 6 , chi= nan , Lambda= inf
iter: 7 , chi= nan , Lambda= inf
iter: 8 , chi= nan , Lambda= inf
iter: 9 , chi= nan , Lambda= inf
problem solve cost: 444.026000 ms
   makeHessian cost: 15.813000 ms
----------- update bprior -------------
             before: nan
                     nan
             after: nan
                    nan
edge factor cnt: 649
1 BackEnd processImage dt: 481.907000 stamp: 1403638526.327830 p_wi: -0.437289 -0.153459 -0.781337
iter: 0 , chi= nan , Lambda= 500000.000000
iter: 1 , chi= nan , Lambda= 18014398509481986097152.000000
iter: 2 , chi= nan , Lambda= 822752278660603116858455895396728899888057129711113907845499648475136.000000
iter: 3 , chi= nan , Lambda= 47634102635436898724379873315377984320228608500171483271481892445135526726758807532347872310828455112538944824267416431834545575140033753564315648.000000
iter: 4 , chi= nan , Lambda= 3495959950985713444632080123730709903643510074804632702858961286653459985776588139814929586748127946306818840941912242619549823910362988237896222302398330053978381405443632220440135622373810791965842646989594435053520339544682443171476172588464958603264.000000
iter: 5 , chi= nan , Lambda= inf
iter: 6 , chi= nan , Lambda= inf
iter: 7 , chi= nan , Lambda= inf
iter: 8 , chi= nan , Lambda= inf
iter: 9 , chi= nan , Lambda= inf
problem solve cost: 419.517000 ms
   makeHessian cost: 15.422000 ms
----------- update bprior -------------
             before: nan
                     nan
             after: nan
                    nan
edge factor cnt: 661
1 BackEnd processImage dt: 457.779000 stamp: 1403638526.427830 p_wi: -0.477645 -0.193317 -0.802495
iter: 0 , chi= nan , Lambda= 500000.000000
iter: 1 , chi= nan , Lambda= 18014398509481986097152.000000
iter: 2 , chi= nan , Lambda= 822752278660603116858455895396728899888057129711113907845499648475136.000000
iter: 3 , chi= nan , Lambda= 47634102635436898724379873315377984320228608500171483271481892445135526726758807532347872310828455112538944824267416431834545575140033753564315648.000000
iter: 4 , chi= nan , Lambda= 3495959950985713444632080123730709903643510074804632702858961286653459985776588139814929586748127946306818840941912242619549823910362988237896222302398330053978381405443632220440135622373810791965842646989594435053520339544682443171476172588464958603264.000000
iter: 5 , chi= nan , Lambda= inf
iter: 6 , chi= nan , Lambda= inf
iter: 7 , chi= nan , Lambda= inf
iter: 8 , chi= nan , Lambda= inf
iter: 9 , chi= nan , Lambda= inf
problem solve cost: 383.488000 ms
   makeHessian cost: 15.482000 ms
----------- update bprior -------------
             before: nan
                     nan
             after: nan
                    nan
edge factor cnt: 668
1 BackEnd processImage dt: 421.356000 stamp: 1403638526.527829 p_wi: -0.520874 -0.237832 -0.827581
iter: 0 , chi= nan , Lambda= 500000.000000
iter: 1 , chi= nan , Lambda= 18014398509481986097152.000000
iter: 2 , chi= nan , Lambda= 822752278660603116858455895396728899888057129711113907845499648475136.000000
iter: 3 , chi= nan , Lambda= 47634102635436898724379873315377984320228608500171483271481892445135526726758807532347872310828455112538944824267416431834545575140033753564315648.000000
iter: 4 , chi= nan , Lambda= 3495959950985713444632080123730709903643510074804632702858961286653459985776588139814929586748127946306818840941912242619549823910362988237896222302398330053978381405443632220440135622373810791965842646989594435053520339544682443171476172588464958603264.000000
iter: 5 , chi= nan , Lambda= inf
iter: 6 , chi= nan , Lambda= inf
iter: 7 , chi= nan , Lambda= inf
2019-11-16 00:19:47.426694+0800 run_euroc[5569:2277701] GetDYLDEntryPointWithImage(/System/Library/Frameworks/AppKit.framework/Versions/Current/AppKit,_NSCreateAppKitServicesMenu) failed.
iter: 8 , chi= nan , Lambda= inf
iter: 9 , chi= nan , Lambda= inf
problem solve cost: 390.331000 ms
   makeHessian cost: 14.197000 ms
----------- update bprior -------------
             before: nan
                     nan
             after: nan
                    nan

Process finished with exit code 9
@HeYijia
Copy link
Owner

HeYijia commented Nov 16, 2019

mac 版本我还真没测试过，不过这个版本应该有两个同学跑通过。看你 log 前面初始化的时候Initialization finish! 前面那一段好像很多 imu 数据丢失。应该是系统没有初始化好（数据有丢失？）就开始跑后面的 vio，所以没多久就跪了。

@BigDeviltjj
Copy link
Author

@HeYijia 所以这个wait for imu, only should happen at the beginning sum_of_wait: 80 其实是不正常的是吧，我debug一下看看，谢谢！

@fu1899
Copy link

fu1899 commented Nov 27, 2019

测试此程序时也遇到类似的现象，定位了根因，并解决了问题。
【测试环境】：ubuntu 14.04
【问题描述】：初始化过程中不收敛，优化时LM会挂掉，现场一致。
【根因描述】：
1、ceres的库引起，构建时因为配置原因，ceres的ceres::DENSE_SCHUR模式无法使用，而初始化流程中SFM部分，默认是使用此方式。
2、构建ceres，进行cmake ..时报以下语句：
....

Disabling the use of Eigen as a sparse linear algebra library.
This does not affect the covariance estimation algorithm
which can still use the EIGEN_SPARSE_QR algorithm.

....
3、与系统版本没有关系，测试使用的是ubuntu 14.04
【解决办法】
方法1、重新构建ceres。（我没尝试）
方法2、将优化的配置参数设置为ceres::SPARSE_SCHUR
代码位置：initial_sfm.cpp文件中第275行options.linear_solver_type = ceres::DENSE_SCHUR;

目前测试时的问题已经解决，你试试是不是同样的原因。

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
3 participants
@HeYijia @fu1899 @BigDeviltjj