Behaviour Trees

Behaviour Trees

Behaviour trees are used to program AI in the game. Anything autonomous from NPCs to enemies are controlled by behaviour trees.

1. Node Types

Conceptually, there are three types of nodes in our behaviour tree model:

1. Input Node
   • Input nodes take data from the world and store them in the behaviour tree. Example: Finding the player entity.
2. Processing Node
   • Processing nodes take stored data and interpret and change that data. These are the core of behaviour trees. Example: Taking the already retrieved player and getting its position.
3. Action Node
   • Action nodes use the data from the input & processing nodes to take certain actions. Example: Moving towards the player's position.

2. Node Documentation

Name	Type	Function
FindActorByTagNode	Input	Finds an actor via the given tag.
GetActorPositionNode	Processing	Gets the position of a stored actor.
GetRandomPositionNode	Processing	Generates a random position within a certain radius from a stored origin.
HasPropertyNode	Processing	Returns success if the agent has the specified property.
NotNode	Processing	Succeeds if its children fail, and fails if its children succeed.
RangeCheckNode	Processing	Succeeds if the agent is within the specified distance of a stored position; else fails.
SelectorNode	Processing	Executes children until one doesn't fail.
SequenceNode	Processing	Executes children until one doesn't succeed.
SimpleMoveNode	Action	Moves the agent toward a stored position ignoring physics and without pathfinding.
SuccessNode	Processing	Always succeeds.

3. Node Methods (IBehaviourTreeNode)

Behaviour tree nodes must inherit the IBehaviourTreeNode interface, which comes with two methods:

void Init(Behaviour behaviour)

The init method is used to create and initialize methods on this node.

NodeStatus Tick(Tree<Behaviour>.Node self, BehaviourObject obj)

Tick is called once per frame when this node is running. This is where a node will process data and do what it needs to do during runtime.