Managing State with Blackboard Systems

In this lesson, you will learn how to use a centralized Blackboard to share data between Behavior Tree nodes, allowing your AI to "remember" information across complex decision paths.

Why Do We Need a Blackboard?

As your Behavior Tree grows, individual nodes often need to share information—like the "last seen position" of a player or whether the AI is currently "low on health." If you pass this data manually through every node, your tree becomes tightly coupled and difficult to maintain. A Blackboard acts as a shared memory container (a simple object or dictionary) that all nodes in a tree can read from or write to.

Implementing the Shared Data Store

Think of the Blackboard as a specialized script that acts as the "brain's notepad." Instead of nodes storing local variables, they reference this central object to coordinate their behavior.

public class AIBlackboard : MonoBehaviour
{
    // Shared state variables
    public Transform targetPlayer;
    public Vector3 lastKnownPosition;
    public bool isAggressive = false;
}

Connecting Nodes to the Blackboard

When your Behavior Tree processes a branch, you inject this AIBlackboard reference into each node during initialization. A "Sense Player" node might write to lastKnownPosition, while a "Move To Target" node reads that same value to update the NavMeshAgent destination.

public class MoveToTargetNode : ActionNode
{
    private AIBlackboard _blackboard;

    public MoveToTargetNode(AIBlackboard blackboard) => _blackboard = blackboard;

    public override NodeState Evaluate()
    {
        if (_blackboard.targetPlayer != null)
        {
            agent.SetDestination(_blackboard.targetPlayer.position);
            return NodeState.Success;
        }
        return NodeState.Failure;
    }
}

By decoupling the data from the node logic, you create a system where nodes can be swapped or reordered without breaking the flow of information. If the player goes behind a wall, your "Sensor" node simply updates the lastKnownPosition in the Blackboard, and the "Movement" node automatically reacts during the next tick without needing a direct link to the sensor itself.

Key Takeaway: A Blackboard system decouples AI logic from data storage, allowing nodes to act as independent modules that communicate through a shared, centralized state container.