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.

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

Baseline results mismatch #137

Closed
Chiasung opened this issue Oct 27, 2017 · 1 comment
Closed

Baseline results mismatch #137

Chiasung opened this issue Oct 27, 2017 · 1 comment
Labels
user-platform User has trouble running on their own dataset

Comments

@Chiasung
Copy link

Hello All,

I am new to Kalibr. Yesterday, I ran two different calibration methods to calibrate my stereo camera with built-in IMU using the same bag file.
The outcome of camera baselines are differnt. One estimation by kalibr_calibrate_cameras is close to the camera spec (6cm), however, the result from kalibr_calibrate_imu_camera is 11cm. That is not reasonable.
Could anybody give a hint to sovle this issue. I will greatly appricate. Following is the commands I used and resuts.

  1. kalibr_calibrate_cameras --bag 2017-10-25-23-07-52.bag --bag-from-to 4 9 --topics /stereo/left/image_raw /stereo/right/image_raw --models radtan radtan --target target.yaml
  2. kalibr_calibrate_imu_camera --target target.yaml --cam camchain.yaml --imu imu0.yaml --bag 2017-10-25-23-07-52.bag --bag-from-to 4 9

1)"kalibr_calibrate_cameras" Calibration results

Camera-system parameters:

cam0 (/stereo/left/image_raw):
 type: <class 'aslam_cv.libaslam_cv_python.DistortedPinholeCameraGeometry'>
 distortion: [ 0.08982447 -0.10728168  0.0010053  -0.0015931 ] +- [ 0.00424448  0.01826095  0.00037677  0.00043009]
 projection: [ 753.53041302  752.68502831  305.75204258  259.78825978] +- [ 0.08965215  0.0831327   0.68836144  0.614528  ]
 reprojection error: [-0.000037, 0.000009] +- [0.167784, 0.161655]

cam1 (/stereo/right/image_raw):
 type: <class 'aslam_cv.libaslam_cv_python.DistortedPinholeCameraGeometry'>
 distortion: [ 0.08490426 -0.10763327 -0.00053635 -0.0014541 ] +- [ 0.00490586  0.02443987  0.00039266  0.00042505]
 projection: [ 756.7709099   756.05422708  304.3187416   252.66928682] +- [ 0.0800266   0.0810337   0.70287441  0.58984044]
 reprojection error: [-0.000030, 0.000016] +- [0.176381, 0.164031]

baseline T_1_0:
 q: [ 0.00059181 -0.00114144 -0.00000764  0.99999917] +- [ 0.00141457  0.00173934  0.000089  ]
 t: [-0.05934035 -0.00044085  0.00079437] +- [ 0.00037338  0.00037877  0.00028811]

    Target configuration:
       Type: aprilgrid
       Tags: 
       Rows: 6
       Cols: 6
       Size: 0.055 [m]
       Spacing 0.0165 [m]

2)"kalibr_calibrate_imu_cameras" Calibration results

Normalized Residuals

Reprojection error (cam0): mean 20.3406304542, median 21.3975997685, std: 4.9424453469
Reprojection error (cam1): mean 20.7587335015, median 20.0094473571, std: 4.26173415853
Gyroscope error (imu0): mean 1.26781864517, median 1.11599698412, std: 0.916665181779
Accelerometer error (imu0): mean 1.12501488855, median 1.09159884222, std: 0.512601978485

Residuals

Reprojection error (cam0) [px]: mean 20.3406304542, median 21.3975997685, std: 4.9424453469
Reprojection error (cam1) [px]: mean 20.7587335015, median 20.0094473571, std: 4.26173415853
Gyroscope error (imu0) [rad/s]: mean 0.0755390138638, median 0.066493194414, std: 0.0546166315968
Accelerometer error (imu0) [m/s^2]: mean 0.134060996163, median 0.130079014676, std: 0.061083575489

Transformation (cam0):

T_ci: (imu0 to cam0):
[[-0.09456731 0.99492168 0.03446566 0.02660472]
[-0.99547097 -0.09484432 0.00648929 0.03579494]
[ 0.00972521 -0.03369589 0.99938481 0.0105476 ]
[ 0. 0. 0. 1. ]]

T_ic: (cam0 to imu0):
[[-0.09456731 -0.99547097 0.00972521 0.03804618]
[ 0.99492168 -0.09484432 -0.03369589 -0.02271926]
[ 0.03446566 0.00648929 0.99938481 -0.01169034]
[ 0. 0. 0. 1. ]]

timeshift cam0 to imu0: [s] (t_imu = t_cam + shift)
0.0

Transformation (cam1):

T_ci: (imu0 to cam1):
[[-0.09667503 0.99456145 0.0387486 -0.08331105]
[-0.99507375 -0.09743712 0.01828228 0.03631671]
[ 0.0219584 -0.03679028 0.99908173 0.00936346]
[ 0. 0. 0. 1. ]]

T_ic: (cam1 to imu0):
[[-0.09667503 -0.99507375 0.0219584 0.0278781 ]
[ 0.99456145 -0.09743712 -0.03679028 0.08674104]
[ 0.0387486 0.01828228 0.99908173 -0.00679062]
[ 0. 0. 0. 1. ]]

timeshift cam1 to imu0: [s] (t_imu = t_cam + shift)
0.0

Baselines:

Baseline (cam0 to cam1):
[[ 0.99998854 0.00216014 0.00427195 -0.11003785]
[-0.00221074 0.99992702 0.01187697 0.00045792]
[-0.00424598 -0.01188627 0.99992034 -0.00064487]
[ 0. 0. 0. 1. ]]
baseline norm: 0.110040692385 [m]

Gravity vector in target coords: [m/s^2]
[ 0.86302785 -9.14407545 -3.43649385]

Calibration configuration

cam0

Camera model: pinhole
Focal length: [711.963634284882, 711.963634284882]
Principal point: [374.86409066481804, 244.78405571004106]
Distortion model: radtan
Distortion coefficients: [0.07888409273935, -0.11353901155482328, -0.0004560888399720081, 8.736957158911675e-05]
Type: aprilgrid
Tags:
Rows: 6
Cols: 6
Size: 0.055 [m]
Spacing 0.0165 [m]

cam1

Camera model: pinhole
Focal length: [711.963634284882, 711.963634284882]
Principal point: [372.5024610181662, 236.71572961443084]
Distortion model: radtan
Distortion coefficients: [0.08185918506773818, -0.12296212768690189, -0.0022227781335232093, 0.0006978538097984978]
Type: aprilgrid
Tags:
Rows: 6
Cols: 6
Size: 0.055 [m]
Spacing 0.0165 [m]

IMU configuration

IMU0:

Model: calibrated
Update rate: 142.0
Accelerometer:
Noise density: 0.01
Noise density (discrete): 0.119163752878
Random walk: 0.0002
Gyroscope:
Noise density: 0.005
Noise density (discrete): 0.0595818764391
Random walk: 4e-06
T_i_b
[[ 1. 0. 0. 0.]
[ 0. 1. 0. 0.]
[ 0. 0. 1. 0.]
[ 0. 0. 0. 1.]]
time offset with respect to IMU0: 0.0 [s]

@xukuanHIT
Copy link

Mark.
I have the same problem, and this issue may help you : #101

@goldbattle goldbattle added the user-platform User has trouble running on their own dataset label Dec 6, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
user-platform User has trouble running on their own dataset
Projects
None yet
Development

No branches or pull requests

3 participants