I know from previous game design experience that I would use ray casting here to detect objects in the world around me (as a player). So that immediately makes it an ability of the player or pawn in our case. So in this scenario I would add a component to the pawn that detects objects using ray casting and highlights them in some way. That would then allow us, perhaps using a mouse button click, to attach the desired object to the pawn and make it a child of the pawn, as far as positioning is concerned, so the grabbed object moves with the pawn. Once in the desired position a further mouse button click could release the object.
Will be interested to see how you tackle this Ben as there are usually many solutions to any given task.