Zombie Survival AI

This was an exam for a course focused on game AI implementation. The task was to make an AI for the human agent. You have to take care of energy, health, stamina, exploration, inventory management and danger avoidance (zombies & purge zones)

I opted for a behavior tree over a finite state machine as I find it easier to understand and debug. I also found it easier to iterate as changing behavior can be as simple as changing the order of some blocks. Although, I did also find some downsides, the fact that it requires quite a bit of memory, doesn't really have memory out of the box and that it often can rerun a bunch of conditions for no good reason.

Looking back, these downsides can of course be solved, and some I did solve at the time. I am still a fan of behavior trees if it comes to AI, however, a finite state machine is quicker to whip up and get running.

Another point of reflection is that I should've spent more time on exploration, danger handling and searching items and less on efficient inventory management. But I am happy with this being one of my first experiences in making game AI.