#1-2 Big Game Project | Behaviour Trees vs Finite Statemachines

So I’ve started a new game project, the game is developed during the course Big Game Project at my university and I’m developing the game together with five other team members.

The first week started off with us creating a backlog and setting up a new Unity 5.0 project and linking it to a new repository. We decided what we wanted to prototype first and came to realize that in order to prototype anything we would need to implement behaviour trees in our game.
Our game is played similar to an RTS. You select settlers and order them to do different stuff such as gathering or combat. I thought that in order to test these actions we had to implement the behaviour trees first. We could have a combat behaviour, a gathering behaviour and more. After implementing a basic version of behaviour trees and testing it out I realized that behaviour trees maybe wasn’t the right choice for us. Behaviour trees are more or less a complex AI. This is not really what we want because we are constantly ordering our settlers what to do, they don’t really have to think for themselves.

After some discussion we came to the conclusion that what we needed in our game was finite statemachines. It felt like I had wasted too much time on behaviour trees to just let them. I took the day after our discussion to implement statemachines, I redid what we currently had in behaviour tree form to statemachine form.

Now that we have finite statemachines instead it feels like the right choice, we can implement some features more simply and I think this will save us some time in the future.

At the end of week two we had a working prototype which included gathering, combat and some environment. We did a small playtest and tried to figure out what to change and complete for week three which i will write about next week.

About Semih Parlayan

2014  Programming

View all posts by Semih Parlayan →