beet

Beet is behavior expressed as entity trees, using Observers for control flow. The entity-based approach is very flexible and allows for multiple behavior paradigms to be used together as needed.

Currently implemented paradigms:

• Behavior Trees
• Basic Utility AI
• LLM Sentence Similarity
• Reinforcement Learning

Hello World

// A demonstration of Fallback control flow
world.spawn(FallbackFlow)
  .with_children(|parent| {
    parent.spawn((
      LogOnRun::("Hello"),
      EndOnRun::failure(),
    ));
    parent.spawn((
      LogOnRun::("World"),
      EndOnRun::success(),
    ));
  })

Examples

⚠️⚠️⚠️ If you'd like to check out this repo please use the stable v0.0.3 commit ⚠️⚠️⚠️

Beet and my other crates it depends on are currently on a scene serialization bugfix Bevy fork, see this issue for details. The fix is scheduled for the 0.14.2 milestone so fingers crossed we'll be back on bevy main from then.

The examples for beet are scene-based, meaning each example provides a scene for a common base app. As Bevy scene workflows are a wip, there are a few Placeholder types used for not-yet-serializable types like cameras, asset handles etc.

Most examples rely on assets that can be downloaded with the following commands, or manually from here.

curl -o ./assets.tar.gz https://storage.googleapis.com/beet-misc/assets.tar.gz
tar -xzvf ./assets.tar.gz
rm ./assets.tar.gz

Bevy Versions