Object Recognition Kitchen

The Object Recognition Kitchen (ORK) is a project started at Willow Garage for object recognition.

There is currently no unique method to perform object recognition. Objects can be textured, non textured, transparent, articulated, etc. For this reason, the Object Recognition Kitchen was designed to easily develop and run simultaneously several object recognition techniques. In short, ORK takes care of all the non-vision aspects of the problem for you (database management, inputs/outputs handling, robot/ROS integration ...) and eases reuse of code.

ORK is built on top of ecto which is a lightweight hybrid C++/Python framework for organizing computations as directed acyclic graphs.

Install

Well, all good things must start so check out the Install.

Quickguide

We know you can’t wait; if you don’t care about the intricacies and want to have a quick overview, follow the Quick Guide

Tutorials

Ok, now that you know a little, you can follow the Tutorials.

General Usage

Now that you have a bit more time, we suggest you read about the Infrastructure to understand how to interact with ORK. You can then go through the different steps of object recognition:

ROS integration

The recognition kitchen was built in a ROS agnostic way, but ROS components were also developed for integration with the ROS ecosystem (e.g. publishers, subscribers, actionlib server, RViz plugin ...). For more info, check out the ROS Integration.

Recognition Pipelines

Several object recognition pipelines have been implemented for this framework. Their documentation is work in progress :) :

Techniques 2D/3D Types of object Limitations
LINE-MOD 2D and/or 3D
  • rigid, Lambertian
  • does not work with partial occlusions
tabletop 3D
  • rigid, Lambertian
  • rotationally symmetric
  • also finds planar surfaces
  • scales linearly with the number of objects
  • the object is assumed to be on a table with no 3d rotation
TOD 2D and 3D
  • rigid, Lambertian
  • textured
  
transparent objects 2D and 3D
  • rigid and transparent
  • Training has to be done on a painted version of the object

Tools

There are several tools that are used by some of the pipeline and you might need them for your own work or pipelines:

Developers’ corner

You like ORK ? Well you can add any pipeline or database to it. It is fairly simple and modular, just follow the Developer Guide

Contacts

For bug reports and comments, please use the GitHub infrastructure or join us on the Google Group.

If you want to cite this work, please use the BibTeX reference:

@misc{ork_ros,
   Author = {Willow Garage, ROS community},
   Title = "{ORK}: {O}bject {R}ecognition {K}itchen},
   howpublished = {\url{https://github.com/wg-perception/object_recognition_core}}
}