The BlenderRobotDesigner of the Neurorobotics Platform (NRP)

Introduction

Welcome

This is the documentation of the RobotDesigner belonging to the Neurorobotics Platform (NRP) of the Human Brain Project (HBP) developed in the sub-project SP-10 Neurorobotics. It is realized as a plugin for the Blender 3D modeling suite written in the Python language. This document provides information about the plugin, its installation, its features and usage, as well as the programming reference for developers.

About

The NRP RobotDesigner software developed in the scope of the Human Brain Project in the sub-project SP10 Neurorobtics. It is part of the Neurorobotics Platform which aims at providing easy accessible and usable tools to neuro-scientists and neuro-roboticists for simulating robots that are controlled by artificial brains in virtual environments. The RobotDesinger is a modeling tool that allows to generate geometric, kinematic and dynamic models, and sensor placement that can be used in the simulation environment of NRP–or any ROS/Gazebo-based environment. While other services of the NRP are created as web-based services that do not require installation and, therefore, administration expertise, the RobotDesigner is a plugin for the freely available (GPL licensed) Blender 3D modeling suite. This design decision has been made due to three reasons:

  1. The full capabilities of a feature-rich modeling suite such as Blender is an enormous undertaking.
  2. Robot design is a process that requires advanced engineering expertise and is time consuming justifying a more complex installation process and interruption of the web-based workflow.
  3. Blender is freely available for a wide range of platforms and installation is very easy.
  4. It is easily extendible.

Under the hood, the plugin extends the built-in data types and provides an improved and clear interface to the modeling software. Future development plans of the robot designer, however, include provide a simplified web-based interface integrated into the NRP to construct robots from building blocks. Therefore, the code base of the standalone software is planned to be used and will be further maintained.

The RobotDesigner is currently developed and maintained under the lead of the Intelligent Systems and Production Engineering department of the FZI Forschungszentrum Informatik in Karlsruhe.

History

The foundations of the NRP RobotDesigner has been initially developed under the name OpenGRASP RobotEditor from 2008-2012 at the Humanoids and Intelligence Systems Lab (HIS) at the Institute for Anthropomatics and Robotics (IAR) of the Karlsruhe Institute of Technology during the European GRASP project [1] [2] as port of the OpenGRASP software. This simulation environment has been published under the GPL v2 license. This Python plugin restricted to the Blender version 2.49 was mainly developed by Stefan Ulbrich (FZI). One of the most notable features of the RobotEditor was support for the Collada v1.5 data 3D asset exchange format. This file format has large support from the industry including companies such as Daimler and Sony. The RobotEditor was the only freely available tool creating valid models for Collada v1.5 which supported the new description facilities for kinematics and dynamics.

Afterwards, the RobotEditor [3] has been completely rewritten to comply the programming interface of newer Blender versions (>2.69) at the High Perfomance Humanoid Technologies Lab (H2T) at the (IAR) of KIT by the main authors Michael Bechtel and Stefan Ulbrich (FZI). Many new feature such as physical properties, joint dynamics, sensors for matching motion capture data support for the Simox robot simulator [4] have been added to the software. This software is still developed and available under the open source GPL license from here.

he NRP RobotDesigner is a fork of the RobotEditor, which has been chosen as the basis after a comparison to competing projects (e.g., phobos). It will enrich the existing project by components required for the NRP (e.g., for communication/file exchange), additional file formats, support for simulators, an installer and an adapted user interface.

License

Similar to its predecessor the RobotEditor the NRP RobotDesigner is published as open source software under the GPLv2 license.

The NRP RobotDesigner

The Human Brain Project [5]

The European Commission has selected the Human Brain Project as one of its two Flagship projects. Over the course of 10 years, the HBP will create an integrated system of six ICT platforms, one dedicated to Neurorobotics. The Neurorobotics Subproject of the HBP is coordinated by a research group led by Prof. Alois Knoll. The Human Brain Project is part of the FET Flagship Programme, which is a new initiative launched by the European Commission as part of its Future and Emerging Technologies (FET) initiative. The goal is to encourage visionary, “mission-oriented” research with the potential to deliver breakthroughs in information technology with major benefits for European society and industry. The Commission envisages the Flagship program as a highly ambitious initiative involving close collaboration with National and Regional funding agencies, industry and partners from outside the European Union. Read more…

The Neurorobotics Platform [6]

The Neurorobotics sub-project will develop the Neurorobotics Platform which will offer scientists and technology developers a software and hardware infrastructure allowing them to connect pre-validated brain models to detailed simulations of robot bodies and environments and to use the resulting neurorobotic systems in in silico experiments and technology development. The goal for the ramp-up phase (editor’s note: The first two years of the project ending in March 2016) will be to develop the Neurorobotics Platform version 1 which will allow researchers to design and run simple experiments in cognitive neuroscience using simulated robots and simulated environments linked to simplified versions of HBP brain models. The Neurorobotics Platform will include a Robot Designer, an Environment Builder, and a Closed-Loop engine, as well as the Neurorobotics Platform host facility. The Neurorobotics Platform will exploit the 3D modelling capabilities provided by commercial and open source gaming platforms (the choice of platform will be made in the early stage of the project). The HBP Neurorobotics Platform will allow researchers to conduct closed-loop experiments, in which a virtual robot is connected to a brain model, running on the HPC platform or on neuromorphic hardware (editor’s note: E.g., the SpiNNaker boards ). Although the capabilities to model virtual robots and environments already exist, and although various labs have created closed-loop set-ups with simple brain models, the HBP platform will be the first to couple robots to detailed models of the brain. This will make it possible to perform experiments exploring the link between low level brain circuitry and high-level function. The first release of the Neurorobotics Platform will offer extremely simple functionality. The majority of technical development work will focus on the goals fixed for the second release, which will provide a flexible environment in which researchers can perform experiments using simulated robots connected to different classes of brain model – simplified versions of the models provided by the Brain Simulation Platform, high level models coming from the Cognitive Architectures and Theoretical Neuroscience sub-projects.

Installation

Important: You should have a latest blender (version>=2.77) on your computer. Your version of blender should have a built-in python. Install on linux and windows:

  1. Download blender from the dedicated website: https://www.blender.org/download/
  2. Download the “installer.blend” from our github repository: https://github.com/HBPNeurorobotics/BlenderRobotDesigner/blob/master/installer.blend
  3. Launch blender and open the installer.blend file, click “run script” and wait until “RD Installation Done!” appears in your terminal
  4. Relaunch blender, you will find the HBP add-on on the top left of blender GUI (Note: if the HBP tab does not appear in the tool shelf, navigate to File->User Preferences->Add-Ons tab, search for and select the NRP Robot Designer add-on, and click “Save User Settings”)

For further information have a look at the Installation section of the User Manual.

Troubleshooting

If you get some compilation issues with any of Python libraries, you can try downloading precompiled .whl files from here: http://www.lfd.uci.edu/~gohlke/pythonlibs/

Features

  • Installer

    The installer comes in a form of a .blend file that contains an installer script that can be directly executed from within Blender. That way, it can detect the used operating system and Blender version and link the files to the correct location as well as select the correct binaries for the platform. For more information, refer to the documentation.

  • Robot modeling

    The robot designer adds functionality to the Blender software with respect to robotics:

    • Kinematic modeling in a scientific/engineering way (e.g., entering transformations in Denavit-Hartenberg convention)
    • Editing of dynamic properties (center of mass and distribution, friction, etc.)
    • Automatic mesh generation
    • Creation of collision models using geometries with fixed size of vertices and safety distance
    • Convex hull computation
    • Conversion from deformable meshes to rigid bodies. This is useful to transform deformable actors such as those created by MakeHuman into robots. This is used to provide a standard humanoid robot model to the NRP.
    • Generation of links and joint geometries based on the kinematic description (still experimental).
    • Placing of sensors (Note: Currently, this includes cameras only. More to follow on request by the NRP and users)
  • File format and ROS/Gazebo support

    In order to interchange models with the Neurorobotics platform the Robot designer has to support additional file formats.

    In the file section robot models can be imported and exported as single files or zipped packages in the Simulator Description Format (SDF) file format. It will be enriched by additional information tags supported by the Gazebo simulator–especially for supporting a plugin developed for the NRP to include joint controllers directly in the robot description file. This file support relies on the PyXB package–a software that translates XML scheme definitions (XSD) into a Python document object model. Currently, the RobotDesigner supports limited export and import of these files.

  • Plugin Core Framework

    Although the RobotEditor is already feature rich, development is cumbersome due to Blender’s conventions for plugin design. The Plugin Core framework is a python package that abstracts and simplifies many of the boiler plate thus allowing the design of larger applications based on Blender. The RobotEditor, therefore, had to be refactored and rewritten in large parts to comply with this framework.

    Especially the dynamic nature of how functionality is added to Blender makes a modern Python development with IDEs (Integrated Development Environments) such as the excellent PyCharmTM which support code completion and refactoring difficult. By using decorators (PEP 0318) and handlers for Blender Operators and Properties, and extended exception handling and logging, developers can easily create even larger projects comfortably. Integration of external debugging is planned and currently under development. Further, mock ups for the Blender API can be generated and used for code completion.

    Current development focuses on extending the framework to support static type checking using MyPy and code analysis (PyLint) on plugin loading.

    The Plugin Core has an extensive documentation and might be released as a separated project in the future for inclusion in different projects.

  • Documentation and coding standards

    The RobotDesigner comes with extensive documentation in form of a user’s and developers manual which explains all steps necessary to setup and run the software as well on how to extend it and use the Plugin Core in general. The code aims at being well-documented (the original code base is currently in the process of being documented) and to meet coding standards such as PEP 8.

New features

  • Muscle Definition

    We added a new section for muscle definition on the robot model. Exploiting the graphical user interface, one can add muscle pathpoints directly in the scene. All points get listed and can be adapted continuously. As well muscle characteristics such as type, force and length can be changed. On export of a SDF package an additional .osim file with muscle definitions get exported which is compatible with OpenSim. Hereby you can directly simulate the defined muscles in the Neurorobotics Platform.

Planned features

One of the key aspects of the ongoing development is data persistence, that is, the ability to store robot models in different file formats and different storage mechanisms. * Human Brain Project Collaboratory Upload: The zipped Robot model package can be directly uploaded to the Neurorobotics platform to be used in the web-based simulator. As well a robot model can be downloaded from a collaboratory to be modified and adapted in the Blender Robot Designer.

More file formats will be supported upon demand. By the inclusion and abstraction of the PyXB interface XML-based systems can be integrated easily.

Furthermore, the plugin core framework will be separated from the main project and will be extended to a web application that allows running a sub-set of Blender’s design capabilities in a web browser. Work on this has been already initiated.

The installer will be improved to install external dependencies automatically for the installed Blender version.


Footnotes

[1]Funded by the European Commission through its Cognition Unit under the Information Society Technologies of the seventh Framework Programme (FP7)
[2]B. Leon, S. Ulbrich, R. Diankov, G. Puche, M. Przybylski, A. Morales, T. Asfour, S. Moisio, J. Bohg, J. Kuffner and R. Dillmann, “OpenGRASP: A Toolkit for Robot Grasping Simulation”, 2nd International Conference on Simulation, Modeling, and Programming for Autonomous Robots (SIMPAR), November 15, 2010
[3]N. Vahrenkamp, M. Kröhnert, S. Ulbrich, T. Asfour, G. Metta, R. Dillmann and G. Sandini, “Simox: A Robotics Toolbox for Simulation, Motion and Grasp Planning”, International Conference on Intelligent Autonomous Systems (IAS), pp. 585 - 594, 2012
[4]C. Mandery, Ö. Terlemez, M. Do, N. Vahrenkamp and T. Asfour, “The KIT Whole-Body Human Motion Database”, International Conference on Advanced Robotics (ICAR), pp. 0 - 0, July, 2015
[5]From the project’s website
[6]From the Neurorobotics website

Indices and tables