π Behaviour Tree addon for Godot Engine!
- Download this repository
- Unpack the
beehave
folder into your/addons
folder within the Godot project - Enable this addon within the Godot settings:
Project > Project Settings > Plugins
Behaviour trees are a modular way to build AI logic for your game. For simple AI, behaviour trees are definitely overkill, however, for more complex AI interactions, behaviour trees can help you to better manage changes and re-use logic across all NPCs.
In a nutshell, a behaviour tree is a Godot Node that can be added as a child to other Godot nodes within the scene tree. It will run its logic every frame tick and modify the parent node accordingly.
In more theoretical terms, a behaviour tree consists of so called nodes - each node can be of a different type with different purposes. Those are described further down below in more detail. Every node has a tick(actor, blackboard)
method that can be used to execute custom logic. When the tick
function is called, beehave expects a return status of either SUCCESS
, RUNNING
or FAILURE
.
Conditions are leaf nodes of type ConditionLeaf
. They should be kept simple and either return SUCCESS
or FAILURE
depending on a single condition. Avoid creating conditions that check multiple things as it will become more difficult to reuse these nodes.
1. Example Condition code: IsVisibleCondition.gd
class_name IsVisibleCondition
extends ConditionLeaf
func tick(actor, blackboard):
if actor.visible:
return SUCCESS
return FAILURE
Actions are leaf nodes of type ActionLeaf
. They can be long running potentially being called across multiple frame executions. In this case return the code RUNNING
.
2. Example Condition code: MakeVisibleAction.gd
class_name MakeVisibleAction
extends ActionLeaf
func tick(actor, blackboard):
if actor.visible:
return FAILURE
actor.visible = true
return SUCCESS
The blackboard is an object that can be used to store and access data between multiple nodes.
In order to create logic flows based on conditions and actions, we need to compose them through so called composites. A composite is a node that executes its children in a particular manner as described below.
Selector nodes will attempt to execute each of its children and reports SUCCESS
status code in case one of the children reports a SUCCESS
status code. In case all children report a FAILURE
status code, this node will also return FAILURE
status code. This node will attempt to process all its children every single tick, even if one of them is currently RUNNING
already.
The Selector Star node is similar to the Selector, however, it will skip all previous child nodes that were executed prior, in case one of the children is currently in RUNNING
state. A usecase for this is if you want to ensure that only one action is executed at a time, regardless of for long it runs.
Sequence nodes will attempt to execute all of its children and reports SUCCESS
status code in case all of the children report a SUCCESS
status code. In case at least one child reports a FAILURE
status code, this node will also return FAILURE
status code. This node will attempt to process all its children every single tick, even if one of them is currently RUNNING
already.
The Sequence Star node is similar to the Sequence, however, it will skip all previous child nodes that succeeded prior, in case one of the children is currently in RUNNING
state. A usecase for this is if you want to ensure that only one action is executed at a time, regardless of for long it runs.
Decorators are nodes that can be used in combination with any other node described above.
A failer node will always return a FAILURE
status code.
A succeeder node will always return a SUCCESS
status code.
A inverter will return FAILURE
in case its child returns a SUCCESS
status code or SUCCESS
in case its child returns a FAILURE
status code.
The limiter will execute its child x
amount of times. When the number of maximum ticks is reached, it will return a FAILURE
status code.
I have recorded this tutorial to show in more depth how to use this addon:
This project has been inspired by the behaviour tree example of https://github.com/viniciusgerevini. In case you want to suggest improvements to this addon or fix issues, feel free to raise a pull request or raise an issue!