Skip to content

Latest commit

 

History

History
119 lines (108 loc) · 5.55 KB

README.md

File metadata and controls

119 lines (108 loc) · 5.55 KB
Raw

The README intstuctions and the tutorials are still under construction

Failure Recovery with Ontologically Generated Behaviour Trees (FROGBT)

Welcome to the Failure Recovery with Ontologically Generated Behaviour Trees (FROGBT) project repository! This project is part of the Master's program in Robotics at the Technical University of Delft and represents a significant step forward in the field of robotics, specifically in the domain of behaviour trees and failure recovery.

Introduction

FROGBT represents a novel fusion of behaviour trees (BTs), a hierarchical task execution method, and ontological reasoning, a framework for structuring knowledge. By seamlessly integrating ontological reasoning into the process of BT generation, FROGBT aims to bridge the gap between generated BTs and robust failure recovery strategies. This integration imbues BT generation with contextual awareness, enabling robots to make informed decisions in complex and dynamic environments.

Getting started

pip install owlready2
pip install lxml
  • You also need the BehaviorTreeCPP package v3.8 which can be installed using the following line
  sudo apt-get install ros-$ROS_DISTRO-behaviortree-cpp-v3
  • Clone BehaviorTrees.ROS repo, a useful package to use actions and services of ROS in BehaviorTreesCPP package.
git clone https://github.com/BehaviorTree/BehaviorTree.ROS.git
  • For visualization you will need GROOT installation instruction
    • GROOT proved somewhat challenging to get working with v3.8 of BehaviorTreeCPP package, thus a fork was made with changes that made it work for my setup.
     git clone --branch wisjaber-patch-1 https://github.com/wisjaber/Groot.git

Since GROOT is only for visualization, if it doesn't work you can comment out the following line in the "bt_generation_action.h" and "tutorial_bt_generation_action.h" files

	PublisherZMQ publisher_zmq(tree_,200);

  • For the skills and simulated environment you need to request access to the gits from AIRLab

  • Clone the Gazebo assets repo. (don't forget to source the environment variables as it says in the github)

     git clone https://github.com/tud-airlab/airlab_gazebo_assets.git

  • Clone Albert skills repo

     git clone https://github.com/tud-airlab/albert_skills.git

  • Clone Albert repo

     git clone https://github.com/tud-airlab/albert.git

  • Clone the msgs package

git clone https://github.com/wisjaber/FROGBT_msgs.git
  • Clone the extra skills for recovery package
git clone https://github.com/wisjaber/airlab_extra_skills.git
  • Clone the main repo
git clone https://github.com/wisjaber/FROGBT.git

How To Use

There are three scenarios to evaluate FROGBT you can run each following the instructions below

Scenario one

This scenario is comparison with SkiROS2 framework in their example task found in SkiROS examples In order to run the scenario using FROGBT build your workspace and source it.

  • Then in a sourced terminal
cd workspace/src/FROGBT
make setup_scenario1
  • Start GROOT, if you have it, and choose monitor alt text
  • In another sourced terminal run
make scenario1
  • Immediately connect to GROOT to see the BT growing. Alternatively, you can see the tree execution in the first terminal

alt text

Scenario Two

This scenario is preparing a list of products in the simulated environment. There are two different ways to evaluate it: Sequence List, and BT Reuse.

Sequence List

  • To start the simulation in a sourced terminal run
cd workspace/src/FROGBT
make simulation
  • sometimes the arm gets into singularity when loading the simulation. If that happened, run the following in a terminal

alt text

make arm
  • Run the servers in a sourced terminal by running
make servers
  • Start the server for the BT execution a sourced terminal by running
make bt_execution
  • Start your GROOT and in a terminal run the following to start the list of products tree generation
make list_bt

BT Reuse

Follow the same steps from the Sequence List instruction behalf the last one. instead run the following

make frogbt

This will run the frogbt main script to place one product, tag=1, thus generating a BT for it. After it is done to reuse that BT run the following

rosrun bt_tests bt_reuse.py place milk <tag>

Replace with the number of the product you want, in the ontology there are 4 products defined with the tags: 1,18,19,20. you can reuse it for a different product multiple times until there are no products defined in the ontology.