The Human Brain Project

Gaining fundamental insights about how the human brain works and, ultimately, what "makes us human" remains one of the greatest challenges of the 21st century. This understanding is exactly the main goal of the large-scale "Human Brain Project" (HBP), a Future and Emerging Technologies (FET) flagship project funded by the European Commission. In this project, over 120 institutions out of 24 countries cooperate in integrating neuro-scientific data using methods from many different disciplines such as medicine, Information and Computer Technologies (ICT), and neuroscience, to achieve a deeper understanding of the human brain. Eventually, these results will pave the way to new treatments for brain diseases and the development of new biologically inspired technologies like neuromorphic hardware and robots with human-like behavior.

In the scope of the HBP, the FZI Research Center for Information Technology at the Karlsruhe Institute of Technology, with the research divisions IDS and SE involved, is one of the partners of the Neuro-Robotics subproject (SP10). In the first phase of the project, this project focused on the development of a software infrastructure for scientists and developers alike for the creation and execution of reproducible neurorobotic experiments. In later phases of the project, this subproject will dedicate its efforts to the development of new robot technologies inspired by theoretical neuroscience.

SUBPROJECT 10: NEUROROBOTICS

The Neuro-Robotics subproject pursues the goal of providing the scientific community with a common unified platform for the collaborative design of experiments that links principles and data from theoretical neuroscience to realistic simulations of robots and interactive environments. That way, the resulting neuro-robotic systems can be investigated and be evaluated easily in silico such that they can contribute to the development of novel robotic technologies. The platform will also offer access to the computational resources of high-performance computing centers and the specialized neuromorphic hardware architectures that are developed in different subprojects of the human brain project. This will enable researchers to simulate even entire brain models at a very realistic and high level of detail and connect them to simulated robot platforms for control in real-time.

With a team of robotics specialists and years of experience in the creation of simulation environments, the FZI contributes to the development of this Virtual Neuro-Robotic Simulation Platform (Project VINERO). Ultimately, this project will provide the software necessary to run closed-loop simulations with brain models connected to both simulated and real robotic platforms capable of interacting with their environments. 

Architecture

The software consists of following core components:

  • The Robot and Environment Designers allow creating virtual robots and environments based on detailed specifications.
  • The Brain and Body designers offer researchers the possibility to model neural circuits and their connection to a robot platform.
  • Complex experiments can be defined by the Experiment designer.
  • Brain models and robot platforms are connected and simulated by the Closed-Loop Engine.

Research topics at FZI

In next phase of the project, starting in April 2016, the FZI will focus its scientific efforts on sonsorimotor representations and learning, and higher-level reasoning. Therefore, the transfer of established and well understood concepts of traditional machine learning to representations based on realistic spiking neural models will be investigated together with the integration of novel hardware such as the retina-inspired Dynamic Vision Sensor (DVS) camera, motion tracking and real-time capable neuromorphic hardware.

The goal of this research is, on the one hand, the creation of a system that creates a bridge between classical machine learning and neural computing in order to provide a realistic model of human sensorimotor coordination. On the other hand, it will be evaluated how the field of robotics can benefit from such systems with respect to the expected increase in flexibility and fault tolerance.

Ranging from robots that model simple systems, such as the inspection robot KAIRO and the six-legged walking machine LAURON, up to complex robots driven by artificial muscles, the involved robotic platforms will reflect the increasing complexity of the required neural control systems. In addition to that, the transfer to industrial applications with robots connected to dedicated neuromporhic hardware will be investigated. 

Work for KIT Students

Praktikum "Virtual Neurorobotics in the Human Brain Project - Spiking Neuronal Networks in Neuro-Robotics"

The internship offers students the opportunity to get to know the research field of neurorobotics in the context of the "Human Brain Project". In the course of the practical course, the concepts of virtual neurorobotics from the modelling of artificial spiking neural networks to the design of suitable experiments for training and evaluation in a simulation environment will be discussed. A closed-loop experiment, defined in the Neurorobotics Platform (NRP), forms the framework of our NRP Challenge. The experiment takes place in a virtual environment and consists of one (or more) robots with different actuators and sensors. A neural simulator is used for the spiking neural networks. We provide the basic pipeline and you will be divided into teams to solve different tasks. The results will then be written down in English in the form of a paper and presented in a 20-minute presentation. Further information can be found in the praktikum description and at: http://his.anthropomatik.kit.edu/.  If you are interested and have any questions, please contact Camilo Vasquez Tieck

Seminar "Human Brain Project"

This seminar focuses about current research in the field of neurorobotics in the context of the Human Brain Project. The topics offered mainly address bio-analogous spiking neural networks for modeling brain functionality as well as their various applications to robotics–but also related topics. The seminar targets students of computer science, physics, mechanical engineering and electrical engineering. As number of participants is limited, a registration via Email is mandatory. Seminar papers may be written and be presented in English. For further information, please refer to the official description and to http://his.anthropomatik.kit.edu

If you are interested in participation or have questions, please contact Camilo Vasquez Tieck

Theses and Research Assistants

Within the framework of HBP, students can either write their final theses (supervised by KIT) or work as student assistants (prerequisites in each case: sound programming knowledge and experience in the fields of robotics or machine learning). Please note the announcements or contact 

Camilo Vasquez Tieck.</p>

Workshops

Contact

Dr.-Ing. Daniel Reichard

more

Career

Dr.-Ing. Daniel Reichard studierte von 2007 bis 2014 Informatik am Karlsruher Institut für Technologie (KIT). 2017 promovierte er am Institut für Robotik und Anthropomatik (IAR) des KIT über das Thema "Bildbasierte Weichgeweberegistrierung in der Laparoskopie" im Rahmen der Arbeitsgruppe "Computerassistierte Chirurgie".

Seit Januar 2018 betreut Daniel Reichard für den Forschungsbereich Intelligent Systems and Production Engineering (ISPE) das Human Brain Project der europäischen Kommission im Rahmen des "Future and Emerging Technologies (FET) Flagship" Programms.

Contact

Phone: +49 721 9654-238
E-Mail: Daniel.Reichard@dont-want-spam.fzi.de

hide details