Implementation of A Multifaceted Surrogate Model for Search-Based Procedural Content Generation pape

Hi
We wanted to implement A Multifaceted Surrogate Model for Search-Based Procedural Content Generation paper, but we don't know where to start.
We know everything about the paper and pipeline we need to deploy but we are new to unity and we don't know how to use PCG to create levels and character classes that we need in such a scale and how to use CNN in unity.
We Also need to use Evolutionary Algorithms.
Also we need 3d FPS game with different characters and weapons classes.
We would appreciate any help in this journey of ours.
Thank You

 

Hi,
Do you have any experience with procedural generation?
If you are lost at this point, you may be onto too much to chew.
Seems that your team doesn't have any expertise in any of basics elements of the project.
If that is the case, you may want to reevaluate, if you really need such complex mechanics.
And if simplistic well known model and solutions can be implemented instead.

Personally I would lookup, if there are any git repos, or other sources on the net.
You can look into paper references.
But don't expect someone here to give the solution.

 

For implementation of neural networks in Unity you can take a look at this tutorial suggesting the use of Pytorch instead of ML-Agents if you are used to it.
Regarding procedural level generation there are different choises depending on what you are going to do, if you need online resources the wave function collapse algorithm is well covered specially for the two dimensional case.
Character classes generation seems to me covered by the paper and for the rest I second @Antypodish.

PS.
If you need (free) 3d FPS assets you may look for them on the asset store.

 

Thank you for your answer.
I know the basics of Unity and currently I'm learning more. But this is too much for us and I'm the only developer on team.
Anyways, Thank you for your response.

 

Thank you.