Render scene from script

We are developing an augmented reality game and would like to run nightly tests on a pre-defined sequence of camera positions. From a bash/python script we would like to call a unity app, which would render a series of video frames given a specified sequence of camera poses. The input looks like

camera_pose1
...
camera_poseN

Where for example the camera poses could be a rotation matrix and a translation vector. From this input we would like to render a series of images corresponding to a particular unity scene/model viewed from the given position, and output to a video file. Any suggestions on how to do this?

You could let the Unity app read values from a text file which you change from a bash or python before it launches. Then have a couple of functions which sets up your scene in a predefined pattern and render out a batch of images.

You could then merge these images into a sequence with an external software.