systems.md

---
permalink: /systems/
layout: single
title:  "Systems"
excerpt: "ToBi Systems"
header:
  overlay_image: /assets/images/home_banner.jpg
classes: wide
gallery:
  - url: /assets/images/system/mtc_approach.jpg
    image_path: /assets/images/system/mtc_approach_th.jpg
    title: "Grasp Task"
  - url: /assets/images/system/objrec-rviz.jpg
    image_path: /assets/images/system/objrec-rviz_th.jpg
    title: "Object Recognition visualization"
sidebar:
  - 
    image: /assets/images/system/ma-controller4x.gif
    image_alt: "Robocup at Home logo"
  - title: ""
    text: |
      - [Platforms](#platforms)
---

Our service robots employ distributed systems with multiple clients sharing information over network. These clients host numerous software components written in different programming languages.

We provide a full specification of the system in our [online catalog platform](https://citkat-citec.bob.ci.cit-ec.net/distribution/tiago-noetic-nightly.xml)

### Reuseable Behavior Modeling

For modeling the robot behavior in a flexible manner ToBI uses the [BonSAI](https://github.com/CentralLabFacilities/bonsai) framework. It is a domain-specific library that builds up on the concept of sensors and actuators that allow the linking of perception to action.

### Development and Deployment Tool-Chain

The software dependencies — from operating system dependencies to intercomponent relations — are completely modeled in the description of a system distribution which consists of a collection of so called recipes. In order to foster reproducibility, traceability, and potential software (component) re-use of the ToBI system, we provide a full specification of the different systems in our [online catalog platform](https://citkat-citec.bob.ci.cit-ec.net/browse/distribution/).

{% include figure image_path="/assets/images/system/CITK.png" alt="this is a placeholder image" caption="Cognitive Interaction Toolkit" %}

### Object Recognition and Manipulation

Our current object recognition is based on [YoloX](https://github.com/CentralLabFacilities/clf_object_recognition/tree/yolox). We augment the 2D recognition results with 3D segmentation and superquadratic fitting of object primitives. For manipulation ToBi utilizes the Task Constructor Framework for MoveIt!, which provides a way to solve manipulation tasks by defining multiple interdependent subtasks.



{% include gallery caption="Object Recognition visualization and resulting object primitives for grasp generation" id="gallery"%}


<!--
## Videos

<video width="100%" controls>
    <source src="{{ site.baseurl }}/assets/videos/open.webm" type="video/webm">
    Your browser does not support the video tag.
</video> 

<video width="100%" controls>
    <source src="{{ site.baseurl }}/assets/videos/rc.webm" type="video/webm">
    Your browser does not support the video tag.
</video> 

<video width="100%" controls>
    <source src="{{ site.baseurl }}/assets/videos/tiago_clf.webm" type="video/webm">
    Your browser does not support the video tag.
</video> 


<video width="100%" controls>
    <source src="{{ site.baseurl }}/assets/videos/tiago_clf2.webm" type="video/webm">
    Your browser does not support the video tag.
</video> 
-->


# Platforms

<div class="entries-{{ entries_layout }}">
  {% include documents-collection.html collection="platforms" sort_by=page.sort_by sort_order="reverse" type="grid" %}
</div>