Das Projekt FI-NEXT setzt eine Reihe ASR-Forschungsaktivitäten im Rahmen von FIWARE fort. FI-NEXT konzentriert sich nun auf die Vereinheitlichung von Schnittstellen und Datenmodellen, sowie die Optimierung der Kommunikation zwischen den verschiedenen Generic-Enabler, um die Ergebnisse aus FI-WARE und FI-CORE zu einer einheitlichen Open-Source-Infrastruktur zu erheben. In FI-NEXT wird sich der Fachbereich ASR auf Linked-Data als aufstrebenden Mechanismus für leistungsfähigere und flexiblere Schnittstellen konzentrieren. Dabei werden sowohl die Schnittstellen zu Diensten, als auch das für die Übertragung von Daten verwendete Modelle in einer standardisierten Sprache semantisch beschrieben. Das Ziel ist, durch die semantische Beschreibung von Diensten und Schnittstellen, das Design und die Bereitstellung von verteilten Anwendungen im Kontext von FIWARE weiter zu erleichtern. Die Arbeit in FI-NEXT steht in direktem Bezug zu Arbeiten im Advanced Web-based User Interface Chapter in FIWARE, z.B. durch die Weiterentwicklung des in FI-CORE entwickelten, neuartigen Synchronization-Generic-Enabler (FiVES) oder auch XML3D als 3D-User Interface GE. Auch Ergebnisse aus dem Projekt ARVIDA fließen in Form der oben genannten semantischen Dienstebeschreibungen unmittelbar in FI-NEXT mit ein. Ziel ist, die Entwicklung einer end-to-end Anwendungslösung zu vereinfachen, die es erlaubt, den Bogen von IoT-Sensoren bis hin zu interaktiven Visualisierungen in Apps zu spannen. Die Entwicklung soll dabei sowohl das Design der Anwendung, als auch das Deployment der benötigten Service-Infrastruktur beinhalten. Die Arbeitsgruppe um Prof. Slusallek will sich darüber hinaus, abhängig von Wahlergebnissen in der Open Source Community, weiter als Leiter und Architekt des WebUI Chapters, sowie als Co-Chair des Technical Steering Committee engagieren.
Ansprechpartner: Dipl.-Inf. René Schubotz
HP-DLF: High Performance Deep Learning Framework
The goal of HP-DLF is to provide researchers and developers in the “deep learning” domain an easy access to current and future high-performance computing systems. For this purpose, a new software framework will be developed, which automates the highly complex parallel training of large neural networks on heterogeneous computing clusters. The focus is on scaling and energy efficiency, as well as high portability and user transparency. The goal is to scale the training of networks designed in existing frameworks, without additional user effort, over a three-digit number of compute nodes.
DLBB: Deep Learning Building Blocks
DLBB is a project funded through the Cluster of Excellence on “Multimodal Computing and Interaction” (MMCI). The goal of the project is to research and define high-performance, cross-platform abstractions via meta-programming for deep learning frameworks. This will, in particular, include
- how to describe the basic building blocks in a textbook-like style
- how to combine and optimize a sequence of building blocks, and
- how to run on different hardware (CPU, GPU, etc).
ProThOS: Programmable Taskflow Oriented Operating System
ProThOS is a research project funded by the German Federal Ministry of Education and Research (BMBF) through a directive for funding for “basic research for HPC software in high-performance computing”. Parallelization in the exascale era is a major challenge not only from the perspective of a programming model but also to the execution environment: data dependencies are not recognized correctly, the execution overhead is too large, heterogeneity can not be used, etc. Efforts to address this issue in a smart intermediate layer fail due to the incurred overhead. ProThOS therefore brings programming and execution closer together and bases the data-flow-oriented programming language closely on the execution environment as well as the language constructs to the operating system. The language model remains C/C++ oriented and it will be shown that these principles can be mapped in an efficient way to heterogeneous infrastructures. By integration into the operating system, the execution overhead is drastically reduced. The DFKI researches and develops in ProThOS mainly the programming of such systems and investigates this on the basis of ray tracing and stencil pipelines.
Project website: https://manythreads.github.io/prothos
Metacca: Metaprogramming for Accelerators
Metacca is a research project funded by the German Federal Ministry of Education and Research (BMBF) through a directive for funding for “basic research for HPC software in high-performance computing”. The goal of Metacca is to extend the AnyDSL framework into a homogeneous programming environment for heterogeneous single- and multi-node systems. To this effect, the existing programming language and compiler will be extended by an expressive type system and language features enabling efficient programming of accelerators. Significant aspects of this extension concern the modeling of memory on heterogeneous devices, distribution of data to multiple compute nodes and improving the precision and power of the partial evaluation approach.
Within the project further support for distribution and synchronization for data-parallel programs will be built on top of these language enhancements as a library making use of AnyDSL’s partial evaluation features. Performance models and static analysis tools will be integrated into the AnyDSL tool chain to support development of applications and tuning of parameters. A runtime environment with built-in performance profiling will take care of resource management and system configuration. The resulting framework is evaluated using applications from bioinformatics and ray tracing. The target platforms are single heterogeneous nodes and clusters with several accelerators.
Project website: https://metacca.github.io
The overall goal of REACT is a systematic, safe and validatable approach to developing, training and use of digital reality with the goal to ensure safe and reliable acting autonomous systems – especially in critical situations. In order to reach this goal, we use methods and concepts of machine learning – especially
Deep Learning and (Deep) Reinforcement Learning (RL) – to learn lower-dimensional submodels of the real world. From these submodels we compile (semi) automatically complex, high-dimensional models in order to identify and simulate the entire range of critical situations. By means of digital reality, we virtually synthesize the otherwise missing sensor data of critical situations and train autonomous systems so that they are able to handle critical situations safe and confident. The aim of the project is to enhance the capabilities of autonomous systems. Therefore we continuously and systematically validate and align synthetic data with reality and adapt the models where necessary.
Since March 2016 the research department of Agents an Simulated Reality together with the Innovative Retail Lab and partners from other research institutions and from industry is developing intelligent added value services in the field of building automation.
The Visualization Framework was developed in a cooperation between the University of Stuttgart and the DFKI. The highly customizable web-based open-source framework allows to connect to different web services, and to analyze the data with a large variety of interactive visualizations that are connected via brushing and linking. The web-based applications created with the framework run on multiple devices such as smartphones, tablets, desktop computers and large display walls.
An interactive live demo can be found here: https://github.com/cimplex-project/visualization-framework
A interactive live demo can be found here: https://cimplex-project.github.io/cimplex-globe.
The GLEAMViz web service hosts simulation data in various formats. All of them contain compressed binary data for fast data exchange and that needs to be decoded on the client in order to be processed and visualized. By using parallel data processing the decoding time can be greatly reduced. Thus,the library offers decoders for GLEAMViz from ISI and agent based model datasets from ISI and FBK, including movement data from the DFKI.
We also have developed a configurator application that tremendously reduces the time to configure and deploy a custom version of the visualization framework including all data and simulation services. Based on a web page URL, a user is now able to select, which views, data and simulation services he wants to include in his custom deployment. The server backend then creates a unique executable file depending on the selections, and offers it for downloading. The downloaded executable, based on node.js and Docker, then fully automatically installs all dependencies, including data and simulation services, locally on the client machine. By using Docker, the installation is isolated and does not affect or modify the client host system. The configurator creates executables for Windows, Mac OSX and Linux.
- Title: Bringing of Citizen, model and Data together in Participatory, Interactive Social Exploratories
- Run Time: 01.01.2015 – 31.12.2017
- FET Proactive Global Systems Science (GSS)
- Grant agreement no: 641191
- Deutsches Forschungszentrum für Künstliche Intelligenz GmbH, Germany
- Eidgenössische Technische Hochschule Zürich, Switzerland
- Universität Stuttgart, Germany
- University College London, United Kingdom
- Közép-európai Egyetem (Central European University), Hungary
- Fondazione Istituto per l’Interscambio Scientifico, Italy
- Consiglio Nazionale delle Ricerche, Italy
- Fondazione Bruno Kessler, Italy
- Video Globe
Hybrid and intelligent human-robot collaboration – hybrid teams in versatile, cyber-physical production environments
The aim of the Hybr-iT joint research project funded by the Federal Ministry of Education and Research (BMBF) is to build and test hybrid teams of humans and robots working together with software-based assistance systems in intelligent environments in industrial manufacturing. Based on a holistic approach to the various disciplines of human-robot collaboration, intelligent planning and simulation environments, assistance systems and knowledge-based robotics, workers in the production process are supported by robots in such a way that this intensive human-robot cooperation is convenient, safe and efficient.
Hybr-iT researches and evaluates the components required for planning and optimizing hybrid teams in an industrial context – in terms of their integration in existing IT and production systems and as necessary for their control in a production operation. From an IT perspective, this will involve heavily distributed systems with very heterogeneous subsystems (such as plant and robot controls, safety, logistic, database, assistance, tracking, simulation, and visualization systems), which are implemented together in a comprehensive resource oriented architecture (ROA). ASR contributes to the ROA and develops the simulation environment for hybrid human-robot teams, using AJAN and Motion Synthesis.
The Hybr-iT project is funded by the Federal Ministry of Education and Research.
Ansprechpartner: Ingo Zinnikus