Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/ethz-asl/kalibr/llms.txt

Use this file to discover all available pages before exploring further.

Kalibr writes all output files to the same directory as the input bag file, using the bag file’s base name (without the .bag extension) as a prefix. This prefix is called the bagtag. For example, if you run: 
kalibr_calibrate_cameras --bag /data/recordings/stereo_april.bag ...
The bagtag is /data/recordings/stereo_april, and output files are written as /data/recordings/stereo_april-camchain.yaml, /data/recordings/stereo_april-results-cam.txt, and so on.

​
Camera calibration outputs

These files are produced by kalibr_calibrate_cameras. 
stereo_april-camchain.yaml
stereo_april-results-cam.txt
stereo_april-report-cam.pdf
stereo_april-poses-cam0.csv   # only with --export-poses
​
<bagtag>-camchain.yaml
YAML
Camera chain calibration parameters: intrinsics, distortion, resolution, ROS topic, and inter-camera extrinsics (T_cn_cnm1) for each camera in the chain. This file is the primary input to kalibr_calibrate_imu_camera via --cams. See camchain.yaml for the full field reference.
​
<bagtag>-results-cam.txt
text
Human-readable calibration summary. Contains reprojection error statistics (mean, median, standard deviation) per camera, the optimized intrinsic and distortion parameters, and the inter-camera baseline transforms with their norms. Useful for quickly assessing calibration quality without opening the PDF.
​
<bagtag>-report-cam.pdf
PDF
Multi-page PDF calibration report. Includes:
  • Text summary of calibration results and error statistics
  • Per-camera reprojection error scatter plots (one point per corner detection)
  • Reprojection error histograms
  • Camera rig visualization (for multi-camera setups)
​
<bagtag>-poses-cam0.csv
CSV
Optimized target poses sampled at camera rate. Only written when --export-poses is passed. Columns:
#timestamp, p_RS_R_x [m], p_RS_R_y [m], p_RS_R_z [m], q_RS_w [], q_RS_x [], q_RS_y [], q_RS_z []
1609459200000000000,0.123456,-0.234567,1.345678,0.999123,0.012345,-0.023456,0.001234
Timestamps are in nanoseconds (integer). Position is in meters. Quaternion is in w, x, y, z order.

​
Camera-IMU calibration outputs

These files are produced by kalibr_calibrate_imu_camera. 
stereo_april-camchain-imucam.yaml
stereo_april-imu.yaml
stereo_april-results-imucam.txt
stereo_april-report-imucam.pdf
stereo_april-poses-imucam-imu0.csv   # only with --export-poses
​
<bagtag>-camchain-imucam.yaml
YAML
Extended camera chain that adds IMU extrinsics to every camera entry. Specifically, each camN block gains T_cam_imu (camera-to-IMU transformation) and timeshift_cam_imu (temporal offset). This is the file you pass to VIO frameworks and to the exporter tools. See camchain.yaml for the full field reference.
​
<bagtag>-imu.yaml
YAML
Calibrated IMU parameters grouped by sensor name. The structure wraps the noise parameters from the input imu.yaml files under named keys (imu0, imu1, …) and adds calibrated fields such as T_i_b and time_offset. Example:
imu0:
  T_i_b:
  - [1.0, 0.0, 0.0, 0.0]
  - [0.0, 1.0, 0.0, 0.0]
  - [0.0, 0.0, 1.0, 0.0]
  - [0.0, 0.0, 0.0, 1.0]
  accelerometer_noise_density: 0.006
  accelerometer_random_walk: 0.0002
  gyroscope_noise_density: 0.0004
  gyroscope_random_walk: 4.0e-06
  model: calibrated
  rostopic: /imu0
  time_offset: -8.0e-05
  update_rate: 200.0
​
<bagtag>-results-imucam.txt
text
Human-readable calibration summary for the camera-IMU calibration. Contains:
  • Normalized residuals and raw residuals (mean, median, std) for reprojection, gyroscope, and accelerometer errors
  • Per-camera T_ci (IMU-to-camera) and T_ic (camera-to-IMU) transformation matrices
  • Per-camera time shift values
  • Inter-camera baselines (for multi-camera setups)
  • Gravity vector in target coordinates
  • Calibration configuration (camera models, target parameters, IMU parameters)
​
<bagtag>-report-imucam.pdf
PDF
Multi-page PDF report for the camera-IMU calibration. Includes:
  • Text summary pages (calibration results and configuration)
  • 3D trajectory plot of the estimated IMU poses
  • Per-IMU measurement rate plots
  • Per-IMU acceleration plots and per-axis acceleration error
  • Per-IMU accelerometer bias evolution
  • Per-IMU angular velocity plots and per-axis angular velocity error
  • Per-IMU gyroscope bias evolution
  • Per-camera reprojection error scatter plots
​
<bagtag>-poses-imucam-imu0.csv
CSV
Optimized IMU body poses sampled at IMU measurement times. Only written when --export-poses is passed. Columns:
#timestamp, p_RS_R_x [m], p_RS_R_y [m], p_RS_R_z [m], q_RS_w [], q_RS_x [], q_RS_y [], q_RS_z []
1609459200000000000,0.123456,-0.234567,1.345678,0.999123,0.012345,-0.023456,0.001234
Timestamps are in nanoseconds (integer). Position is in meters. Quaternion is in w, x, y, z order. This format matches the EuRoC/ETH groundtruth CSV convention.

​
Interpreting calibration quality

Good camera calibration typically shows mean reprojection errors below 0.5 px. Errors above 1 px suggest a poor target extraction, an inappropriate camera model, or real optical issues (e.g. strong vignetting or motion blur in the bag). For camera-IMU calibration, inspect the gyroscope and accelerometer residuals in the text report. Large residuals relative to the noise density values in imu.yaml indicate the IMU noise parameters may be underestimated, or that there are synchronization problems in the bag.

Build docs developers (and LLMs) love

Get started for freeTalk to us