Unreal Engine for machine learning research

Dear community,

I am very new to Unreal Engine 4 and I want to use it to make some progress in machine learning research. It seems to be an awsome tool to generate photorealistic environments wich would be great to use for simulation.

To be able to use it I especially need to figure out two main questions:

1) Separate landscape in World Outliner

From the online tutorials i have learned how to create a landscape in UE4 and how to use foilage. The problem is the entire landscape then is referenced by one single name in the “world outliner”. For me it would be very important, that different parts of the landscape are listed under different names in the “world outliner”. This is due to the plugin i am using by microsoft research.
Example: Say i am creating a forest landscape. Parts of it may be grass and other parts of it may be leafs and some dirt tracks. What would be the best way of creating such a landscape but at the same time beeing able to differentiate between all of the differnet parts in the “worl outliner”? Do I have to place the entire grass as a static mesh? (The number of instances a particular part consists of wouldnt matter.) How could I model the road with a static mesh? Or is there a better way than using static meshes?

2) API Moving assets

Is there some kind of api, so that i could move, scale and change assets inside the level i created?
Example: Say i brougth in a couple of tree asstets into my level, which i placed and scaled indepenently. Is there a way, to change their position, scale or appearance via an api? So that i could write a script to make changes within the level.

I would be really thankful if some of you could help to figure out these questons or point me in the right direction.

Thanks a lot in advance.