Kinect calibration is required to improve the accuracy of the Kinect data. In this robot, Kinect is used instead of a laser scanner. We can generate data equivalent to that provided by laser scanner by converting Point Cloud data, using a depth image to laser scanner converter package in ROS. This converted data may not be as precise as an actual laser scanner, so in effect, the error from the converted laser scanner can affect robot mapping, navigation, and localization. To reduce the errors to some extent, we can do a calibration prior to our application. Kinect can even work on factory settings without being calibrated, each device has its own camera parameters and these can change from device to device. Some of the camera parameters are focal length, format size principle point, and lens distortion. When we perform camera calibration, we are able to adjust these values.
One of the calibrations used in Kinect is intrinsic calibration. Some of the...
