Replace Camera Parameters with my Own Calibrated Points

I was analyzing the way in which the Reconstruction.json was constructed and was a bit confused on a few things. I currently am using OpenSfM and a robotic arm with a camera to capture images. The robotic arm has some packages which can capture the camera position at each point which it takes a picture. I would like to use this to get the most accurate camera position at every picture and replace the camera positions in reconstruction.json with the ones from the robotic arms calibration.

Where I am confused is that the reconstruction.json already has points in the points with coordinates, which I assume will be used in the point cloud.

Is there a way where I could first replace all the camera positions with my own coordinates before points are computed? This is so that I wouldn't also have to transform the points in the point cloud from reconstruction.json.

Start by tracing how Reconstruction.json is produced, especially how camera positions and point coordinates are computed. Determine whether calibrated robotic-arm positions can be supplied before point computation without transforming the resulting point cloud; done would require a clearly documented or implemented supported workflow.