The 'Final Year Project' Iterative artificial intelligence algorithm

This project is ambitious to say the very least. Given that I took to C# like a duck to water (having written no code since the days of BASIC no less) I am so eager to explore some of the ideas I have had through the years. This particular endeavour is intended to test if a set of identical non-player characters can 'learn' a world in which they are forced to fight.

Most games increase difficulty by adjusting a character's health, its ability to accurately hit a target and by giving it prior knowledge about the world it will fight in.

I wanted more than anything to see if this 'cheating' could be done away with and have characters improve their own resilience by granting them the advantage of storing and analysing experiences, having to explore the world and discover pickups before they can decide to collect them, for example.

Additionally, information pertinent to each character will be stored in special 'nodes' in the world, so that previous experiences related to each area can be studied before decisions are made. Ultimately, each character should evolve its own perceptions and tactics as a result of this.

These first two videos are using GameObjects as nodes, but I found significant problems using the Unity Navmesh. Even so, I programmed them to disover and explore new areas, and to favour pickup of weapons.

This more recent video shows how the same sort of system works, but using the A* algorithm instead of the NavmeshAgent. I have learnt more about C# during this latest phase than all my previous experience combined, and I am very much looking forwar to what I can bring to the project next.