Origin of camera coordinate system: How to interpret the translation component of the T_ci in calibration report

[Jagatpreet Singh Nir]

Hi 

I used Kalibr for camera IMU calibration. Attached are reports generated by Kalibr-cde package. 

Here is the link of the image for our VIO setup to show our coordinate system.

https://drive.google.com/file/d/1ZPUCtKkYGkvSqBiNBizaW1K3KRENL7a8/view?usp=sharing

The link for reports is 

YAML file : https://drive.google.com/file/d/19TyJG1Ui_Bnw4Uie6mresbD_QPOuoA2q/view?usp=sharing

pdf file with spline error: https://drive.google.com/file/d/1zsXn8dulTaG2EKgEbuv6-qApDYfKDAb9/view?usp=sharing

The camera z is pointing forward. We have calibrated the camera (with blue wire) with IMU(red). The coordinate system of IMU is marked on IMU. We are using a global shutter camera.

I am getting very good results for the orientation matrix(rotation) for IMU to Camera axis as you will be able to see. Both the coordinate frames align in a correct way as seen from the picture. 

However, I am not able to understand the translation component of the calibration T_ci. Where is the origin of the camera frame fixed? Is it at the optical center of the lens? Or is it at the center of the image CMOS sensor?

How do we interpret the camera origin in the whole enclosure of the camera? I was not able to find any reference for the same from the papers mentioned on the Kalibr github repo. 

Please explain.