Camera IMU calibration

Thomas Schneider edited this page Oct 23, 2018 · 28 revisions

The camera-imu calibration tool estimates the spatial and temporal parameters of a camera system with respect to an intrinsically calibrated IMU. Image and IMU data has to be provided in a ROS bag.

The calibration parameters are estimated in a full batch optimization using splines to model the pose of the system. Detailed information about the approach can be found in the following papers: (see 1, 2)

Tutorial: IMU-camera calibration

A video tutorial for the IMU-camera calibration can be found here:

alt text

(Credits: @indigomega)

How to use it

1) Requirements

The intrinsic parameters of the IMU (e.g. scales, axis misalignment, nonlinearities,...) need to be calibrated beforehand and and its correction applied to the raw measurements.

Further an IMU configuration YAML has to be created containing the following statistical properties for the accelerometers and gyroscopes:

  • noise density
  • bias random walk

Please refer to the YAML formats page for the data format.

2) Collect images

Create a ROS bag containing the raw image streams either by directly recording from ROS sensor streams or by using the bagcreater script on a list of image files and a CSV file containing the IMU measurements.

The calibration target is fixed in this calibration and the camera-imu system is moved in front of the target to excite all IMU axes. It is important to ensure good and even illumination of the calibration target and to keep the camera shutter times low to avoid excessive motion blur.

Good results have been obtained by using a camera rate of 20 Hz and an IMU rate of 200 Hz.

Tips:

  • try to excite all IMU axes (rotation and translation)
  • avoid shocks, especially at the beginning/end when you pick up the sensor
  • keep the motion blur low:
    • low shutter times
    • good illumination

WARNING: If you are using a calibration target with symmetries (checkerboard, circlegrid), movements which could lead to flips in the target pose estimates have to be avoided. The use of an Aprilgrid is recommended to avoid this problem entirely.

3) Running the calibration

The tool must be provided with the following input:

  • --bag filename.bag
    ROS bag containing the image and IMU data
  • --cam camchain.yaml
    intrinsic and extrinsic calibration parameters of the camera system. The output of the multiple-camera-calibration tool can be used here. (see YAML formats)
  • --imu imu.yaml
    contains the IMU statistics and the IMU's topic (see YAML formats)
  • --target target.yaml
    the calibration target configuration (see Cailbration targets)

The calibration can be run using:

kalibr_calibrate_imu_camera --bag [filename.bag] --cam [camchain.yaml] --imu [imu.yaml] --target [target.yaml]

The temporal calibration is turned on by default and can be disabled using the --no-time-calibration argument. More information about options is available using the help argument:<br>

kalibr_calibrate_imu_camera --h

4) The output

The calibration will produce the following output files:

  • report-imucam-%BAGNAME%.pdf: Report in PDF format. Contains all plots for documentation.
  • results-imucam-%BAGNAME%.txt: Result summary as a text file.
  • camchain-imucam-%BAGNAME%.yaml: Results in YAML format. This file is based on the input camchain.yaml with added transformations (and optionally time shifts) for all cameras with respect to the IMU. Please check the format on the YAML formats page.

An example using a sample dataset

Download the dataset from the Downloads page and extract it. The archive will contain the bag, calibration target and IMU configuration file.

The calibration can be started with:

kalibr_calibrate_imu_camera --target april_6x6.yaml --cam camchain.yaml --imu imu_adis16448.yaml --bag dynamic.bag --bag-from-to 5 45

NOTE: Because there are shocks in the dataset (sensor pick-up/lay-down), only the data between 5 to 45 s is used.

References

Please cite the appropriate papers when using this toolbox or parts of it in an academic publication.

  1. Paul Furgale, Joern Rehder, Roland Siegwart (2013). Unified Temporal and Spatial Calibration for Multi-Sensor Systems. In Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Tokyo, Japan.
  2. Paul Furgale, T D Barfoot, G Sibley (2012). Continuous-Time Batch Estimation Using Temporal Basis Functions. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), pp. 2088–2095, St. Paul, MN.
You can’t perform that action at this time.
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session.
Press h to open a hovercard with more details.