PYBTS - Python Behavior Tree

Overview

pybts (Python Behavior Tree) is a Python library for creating and managing behavior trees, which are used to model the decision-making process in artificial intelligence systems, such as in games or robotics. The library provides a structured way to organize complex behaviors through a hierarchy of nodes, each representing a specific action, decision, or condition.

Features

Node Hierarchy: Implements various node types such as Action, Composite, Decorator, and Condition, allowing complex behavior modeling.
Extensible Framework: Users can define custom nodes by inheriting from base classes like Node, Composite, or Decorator.
Memory Management: Nodes like Sequence and Selector can have memory, maintaining state between ticks.
Parallel Execution: Supports parallel node execution with customizable policies (SuccessOnOne, SuccessOnAll).
Behavior Tracking: Integrates with a Board class to track and log the state of the tree during execution.
Web Interface: Features a web server (BoardServer) to visualize and manage behavior trees through a web interface, including real-time updates.

Key Components

Node Classes: Define the behavior and structure of the behavior tree.
- Node: Base class for all behavior tree nodes.
- Composite, Decorator, Sequence, Parallel, Selector: Specialized node types for structuring tree logic.
Tree Management
- Tree: Represents the entire behavior tree, initialized with a root node.
- Board: Manages logging and tracking of tree state and history.
Web Server
- BoardServer: Flask-based web server for visualizing and managing behavior trees. Supports dynamic data updates and visualization through ECharts.

Installation

Currently, pybts is not available through package managers and must be installed by cloning the repository:

pip install pybts
pip install pybts[rl] # add reinforcement learning support
# or
git clone https://github.com/wangtong2015/pybts.git
cd pybts
pip install -r requirements.txt
pip install .

Usage

Define Behavior Nodes: Create custom behavior nodes by extending pybts.Node or other specific node types like Action, Condition, etc.
Build the Behavior Tree: Use the pybts.builder.Builder to create trees from your nodes.
Track and Log: Initialize a Board object with your tree to enable tracking and logging.
Visualize and Manage: Start the BoardServer to view and interact with the behavior tree in a web interface.

Example

from pybts import Tree, board, builder
from pybts.node import Action
from py_trees import common
import time


# Define custom behavior node
class Person(Action):
  def __init__(self, name: str, age: int):
    super().__init__(name=name)
    self.age = age

  def update(self) -> common.Status:
    self.age += 1
    return common.Status.SUCCESS


# Build the behavior tree
builder = builder.Builder()
builder.register('Person', Person)
# or
builder.register_node(Person)  # will register tag 'Person' and 'person'(snake case)
root = builder.build_from_file('demos/demo_bt.xml')
tree = Tree(root=root, name='Person')

# Initialize board for tracking
bt_board = board.Board(tree=tree, log_dir='logs')

bt_board.clear()
for i in range(10000):
  tree.tick()
  bt_board.track(info={
    'test_info': i
  })  # track the tree status
  time.sleep(0.5)

Web Interface

Running the Board Server:

Use the following command to start the BoardServer with the specified log directory, enabling debug mode, and setting the host and port:

python -m pybts.board_server --dir=logs --debug --host=localhost --port=10000

This command starts the pybts board server using the logs directory for storing and retrieving behavior tree logs. Debug mode is enabled, and the server is accessible at http://localhost:10000.
Alternative Command: If you have a command line interface setup for pybts as a package, you can also start the server using a more direct command:

pybts --dir=logs --debug --host=localhost --port=10000

This assumes that pybts is configured as a command-line tool that internally calls pybts.board_server.

After running the appropriate command, you can open a web browser and navigate to http://localhost:10000 to view and interact with the behavior tree visualizations and management tools provided by pybts.

Acknowledgements

This project, PYBTS (Python Behavior Tree), is developed based on the py_trees library, a powerful and flexible Python framework for building and managing behavior trees. We extend our heartfelt thanks to the py_trees project and its contributors for providing the foundational components and concepts upon which pybts is built.

For more information about py_trees and to access its source code, visit the official GitHub repository: py_trees on GitHub.

We appreciate the effort and expertise that has gone into py_trees, making it possible for us to develop advanced features in pybts and offer a comprehensive behavior tree solution in the Python ecosystem.

Development and Contributions

Contributions to pybts are welcome! You can contribute by submitting issues, providing updates to documentation, or submitting pull requests with new features or bug fixes.

jayscoder/pybts