Based on the work of (Marzinotto et al., 2014), in which they propose a formal, consistent and general definition of Behavior Trees;
Optimized to control multiple agents: you can use a single behavior tree instance to handle hundreds of agents;
It was designed to load and save trees in a JSON format, in order to use, edit and test it in multiple environments, tools and languages;
A cool visual editor which you can access online;
Several composite, decorator and action nodes available within the library. You still can define your own nodes, including composites and decorators;
Completely free, the core module and the visual editor are all published under the MIT License, which means that you can use them for your open source and commercial projects;
Download the file
behavior3-VERSION.min.js and include it into your HTML:
You can use the online visual editor to design your behavior tree:
It is very recommended that you take a look into the user guide, for both core and editor, where you will find examples and descriptions of features in the library:
You can consult the API in:
This library include the following core structures...
- BehaviorTree: the structure that represents a Behavior Tree;
- Blackboard: represents a "memory" in an agent and is required to to run a
- Composite: base class for all composite nodes;
- Decorator: base class for all decorator nodes;
- Action: base class for all action nodes;
- Condition: base class for all condition nodes;
- Tick: used as container and tracking object through the tree during the tick signal;
- BaseNode: the base class that provide all common node features;
In order to build the library or generate the documentation you must have NodeJS and gulp, and install the proper dependencies:
you get the concatenated and minified version of the library as well the documentation folder and compacted version.
Through the development, you can use
in order to watch the source files and run jshint automatically.