Skip to content

Behavioural Trees in ERGS

Behavioural Trees (BT) are a powerful alternative to standard state-based logic for managing complex automated behaviours, NPC-like interactions, and sophisticated room environmental systems. While standard logic in ERGS is often event-driven, Behavioural Trees are status-driven, constantly evaluating the environment to decide the best course of action.

Core Concepts: Success, Failure, and Running

Every node in a Behavioural Tree returns one of three statuses to its parent:

  1. Success: The task was completed successfully or the condition was met.
  2. Failure: The task could not be completed or the condition was not met.
  3. Running: The task is still in progress (e.g., a "Wait" timer or a moving motor).

The "Logic" (Tri-state) System

To handle advanced escape room scenarios, ERGS introduces the Logic (Status) type, which extends standard Boolean logic with a neutral state: * 1 (Success): Task finished / Active. * 0 (Failure): Task in progress / Error / Failed. * -1 (Neutral): Idle / Not started / Task disabled.

Key Node Types

1. The Entry Point

  • Root: The mandatory starting point. It serves as the single entry point for evaluation. In ERGS, the Root can have a TickInterval (e.g., every 5–10 seconds) to manage "Director" logic without overloading the CPU.

2. Composite Nodes (Decision Makers)

  • Selector: The "Priority" node. It evaluates children from left to right and stops at the first one that succeeds. Use this for fallback logic (e.g., "If emergency, do X; otherwise, do Y").
  • Sequence: The "Checklist" node. It evaluates children in order and stops if any of them fail. Use this for linear multi-part tasks.

3. Decorators (Conditions & Filters)

Decorators are attached to nodes to control their execution flow based on specific conditions.

  • GameState Decorator: Specifically designed for escape rooms. it monitors the status of game checkpoints:
    • Is Finished?: Checks if the task is already completed.
    • Is In Progress?: Checks if the players are currently working on it.
    • Is Late?: Checks if the players have exceeded the Recommended or Late Time defined in the GameState.
  • Blackboard (Global Var) Decorator: Compares values in global variables, such as hint counts or player frustration levels.
  • Inverter: Flips the result (Success becomes Failure and vice versa).

4. Leaf Nodes (The Doers)

  • Action: Triggers a standard ERGS action (e.g., relay, sound, LLM hint) and returns Success or Failure.
  • Wait: Pauses the branch, returning "Running" until the timer expires.

Use Case: The "Game Director"

Behavioural Trees in ERGS act as an automated Game Director, dynamically adapting the experience:

  1. Adaptive Difficulty: If a GameState Decorator returns Is Late, the tree can trigger an Action to simplify the puzzle or provide a hint.
  2. Smart Hinting: The tree can check if a puzzle is In Progress for too long before allowing the LLM NPC to speak.
  3. Auto-Progression: If players are critically behind, the BT can trigger an AutoSolve action to open a door and keep the game moving.
  4. Atmospheric Triggering: If the tree detects an Idle state (no progress for a long time), it can trigger ambient sounds or lighting effects to re-engage players.

Best Practices

  • Left-to-Right Priority: Always place your most critical or "emergency" logic on the leftmost branches of a Selector.
  • Throttle the Root: Use a slower tick rate (e.g., 5s) for high-level "Director" logic to save resources.
  • Visual Debugging: Use the ERGS editor to monitor connection colors—Green for Success, Red for Failure, and animated pulses for Running.