How to: Extrinsic Calibration

As robots frequently tumble and fall down, we need to adjust not correctly aligned parts of the robots in their calibration.

Additionally, we need the Inverse Perspective Mapping (IPM)

to map correctly camera pixels to field coordinates.

In order to adjust the calibration of the visualization you can change the roll (offset_x), pitch (offset_y) and yaw (offset_z) of the camera and the IMU. The camera parameters change the camera direction and the IMU parameters change the orientation of the body.

Setup

Start the visualization launch-file including the necessary motion and vision nodes.

ros2 launch bitbots_extrinsic_calibration viz_extrinsic_calibration.launch

In Dynamic Reconfigure open the parameters (left panel) for the nodes: bitbots_extrinsic_imu_calibration and bitbots_extrinsic_camera_calibration.

Do the calibration

Open config file in bitbots_misc > bitbots_extrinsic_calibration > config
Open rqt and navigate to Plugins > Configurations > Dynamic Reconfigure where you can configure the parameters.

Note

If you change the calibration first change all parameters to 0.0. Then start with the adjustment of the IMU parameters.

This is an interactive process and we might need to do a few alternating steps of imu and camera calibration to get a good solution

IMU Parameters

change if the lines are unequally distanced to the right/left (offset_x) or to the front / sides (offset_y)
has a right-handed coordinate system

../../_images/right_handed_coordinate_system.png

Camera Parameters

change (offset_x) if all lines are too far away or too close
change if the lines are not aligned equally on both sides / front (offset_y) due to the rotation of the head which leads to an error in the cameras frame of reference being present in all directions
has a camera coordinate system

../../_images/camera_coordinate_system.png