Knowledge-based Execution Configuration for Adaptive Behavior Trees

Description

This repository contains the knowledge files supporting the implementation and evaluation of the methods presented in the paper "Knowledge-based Execution Configuration for Adaptive Behavior Trees.", part of the BeAware framework.

The included resources provide essential components to understand and extend the example scenario discussed in the paper, including:

• Behavior Trees (BTs): Templates defining the structure and flow for the example scenario’s adaptive behaviors.

• Ontologies:

  • beaware_ontology.owl: The base ontology to use in the BeAware Framework.

  • application_ontology_example.owl: An instantiated ontology tailored for the specific example scenario.

• PDDL: A Planning Domain Definition Language (PDDL) domain file that formalizes the planning aspects of the example scenario.

Example video

The following video illustrates the execution of two Pick Behaviors, one being the Base Template without Flavors, the other with 5 example Flavors applied.

---

Citation

If you use this work, please cite the paper:

@Article{Ruiz-Celada2026,
author={Ruiz-Celada, Oriol
and Rosell, Jan
and Su{\'a}rez, Ra{\'u}l},
title={Knowledge-based Execution Configuration for Adaptive Behavior Trees},
journal={Journal of Intelligent {\&} Robotic Systems},
year={2026},
month={Mar},
day={06},
abstract={Automated planning is commonly used to obtain plans to solve particular tasks. To execute these plans, Behavior Trees have emerged as a popular execution architecture due to their reactivity and modularity. Configuring the execution of a plan into a Behavior Tree requires expanding the high level actions into the proper Behavior Tree structure. Typically, this is achieved using of pre-defined rigid templates. In this work, we propose a novel approach to generating the Behavior Trees using ontologies. The generated Behavior Trees are tailored to the specific requirements of a task and the world by using modifiers to a base template that provides a general solution to the task. These modifiers and their properties are formally defined using ontologies. A proof of concept is developed, illustrating how the Behavior Trees for the execution of manipulation tasks in a kitchen scenario can be adapted to varying conditions by applying different modifiers.},
issn={1573-0409},
doi={10.1007/s10846-026-02373-1},
url={https://doi.org/10.1007/s10846-026-02373-1}
}

---

Authors

• Oriol Ruiz-Celada (Corresponding Author) Institute of Industrial and Control Engineering
  Universitat Politècnica de Catalunya, Barcelona, Spain
  Email: oriol.ruiz.celada@upc.edu

• Jan Rosell
  Institute of Industrial and Control Engineering
  Universitat Politècnica de Catalunya, Barcelona, Spain
  Email: jan.rosell@upc.edu

• Raúl Suárez
  Institute of Industrial and Control Engineering
  Universitat Politècnica de Catalunya, Barcelona, Spain
  Email: raul.suarez@upc.edu

Correspondence

For any inquiries regarding this repository or the associated paper, please contact the corresponding author: Oriol Ruiz-Celada at oriol.ruiz.celada@upc.edu.