Skip to content

An implementation of the NEAT-Algorithm and an UE4 project to try it out.

License

Notifications You must be signed in to change notification settings

Georg-S/Neuro-Evolution

Repository files navigation

This is the implementation of my Bachelor's thesis project.

Neuro-Evolution

The folder "NeuroEvolution" contains the source code for two neuro evolutionary algorithms.

The first one is called "NEAT" ("NeuroEvolution of Augmented Topologies") and is an Algorithm discovered by Ken Stanley.

The second one is called "HNE" ("Historical Neuro Evolution") and was developed by me.

Unreal Engine Project

This is an application of the NEAT algorithm.

50 agents are trying to reach the goal and must not collide with anything (except the ground and other agents)

The agents are equipped with sensors:

Sensoric

The agents are set to inactive if they collide or fall of the map.

Furthermore, they must progress or they get set to inactive.

If every agent is set to inactive a new iteration starts:

Iteration

Main sources for the devlopment of NEAT

Downloadlink for in depth description of NEAT

Mat Buckland, 2002, AI Techniques For Game Programming

About

An implementation of the NEAT-Algorithm and an UE4 project to try it out.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published