Your thinking seems to be a bit twisted to me (no offense...
) or i just don't get it. So you have some location in space, where this AR-marker is? And you want to place a model at that location with a given orientation? Well, then just translate the model. Object3D has a translate()-method and lets you specify rotations either by using some rotate-methods, by setting a rotation matrix directly or by using the align-method.
There are basically three spaces that are important in this case: Object space, world space and camera space.
Object space is the space, in which the model itself is defined, i.e. the coordinate system used in the modeller. An objects position in object space never changes unless you manipulate an object's vertices directly.
World space represents the world, i.e. the locations of those objects (and camera and lights and...) in a world.
Finally, camera space is the space which is used to render the scene. In camera space, the camera is always located at the origin and the rest of the world gets tranformed accordingly.
An example: An object might have it's center around the origin in object space. For example a simple cube reaching from -10 to 10 on all three axes. It's position in world space might be 100,0,0, i.e. it sits 100 units away from the origin on positive x axis. That's what you do by calling translate(100,0,0) on it. The camera might be located at 50,0,0 (in world space!), so the final position of the cube's center in camera space is 50,0,0.