Applied Artificial Intelligence

Current Research Projects on Artificial Intelligence

Overall 27 research projects of the FZI currently deal with applied artificial intelligence. We present you some of these projects of the FZI here.

AuRorA - Reusable, interactive behaviour for proactive robots in the smart home

Current developments in the context of human-robot cooperation in production, in the fields of manipulation and navigation, open up new opportunities and markets for service robots outside industrial workshops. This project AuRorA concentrates on supporting and relieving people in everyday tasks through interactive and proactive robot behaviour in a smart home or smart hotel.

The robot serves as a real avatar that makes interaction with the abstract environment easier as well as more natural and offers motoric assistance. What is innovative is that the robot independently learns new interactive behaviour that links events such as "ringing at the door" with robot reactions. Depending on the transition probabilities already learned, the robot's behaviour against humans changes from a questioning to an informative character. The adaptive dialogues together with the non-verbal reactions of the robot ensure a high acceptance and quality of the interaction. The learned behaviour is intensively evaluated in two practical scenarios (receiving/guiding guests and cooking) with regard to robustness and transferability.

Facts about the project

  • Technology fields: Language and Text Understanding, Sensors and Communication, Human-Machine Interaction and Assistance Systems, Robotics and Autonomous Systems, Machine Learning
  • Application market/sectors:  Construction and Infrastructure, Health Care and Pharma, other services
  • Application field Industry:  Intelligent Assistance Systems, Robotics
  • Development partners:  Charité – Universitätsmedizin Berlin, Myestro Interactive GmbH, Mojin Robotics GmbH
  • Funding authority:  German Federal Ministry of Education and Research
  • Contact person: Arne Rönnau

CyberProtect

The project CyberProtect, funded by the Baden-Württemberg Ministry of Economic Affairs, Labour and Housing, pursues the objective of better safeguarding complex software systems, in terms of strengthening SMEs in Baden-Württemberg. All three fields of security (security, safety and privacy) are considered here, whereby the focus is set on to the subfield of security.

In the framework of the project, methods are developed, in order to make the behaviour as well as decisions of complex software systems, such as AI systems, visible and thus enable making statements about the systems' state of security. SMEs are integrated into the project by means of a wide-ranging offer of quick checks, trainings and open lab days, in order to provide them with the opportunity of examining their complex software for security and, if necessary, improving it.

Facts about the project

  • Technology fields: Language and Text Understanding, Sensors and Communication, Human-Machine Interaction and Assistance Systems, Robotics and Autonomous Systems, Machine Learning
  • Application market/sectors:  Manufacturing Industry/Production, Logistics
  • Application field Industry:  Robotics, other
  • Development partners:  Fraunhofer Institute of Optronics, System Technologies and Image Exploitation IOSB, Fraunhofer Institute for Manufacturing Engineering and Automation IPA
  • Funding authority:  Ministry of Economic Affairs, Labour and Housing Baden-Württemberg
  • Contact person: Arne Rönnau

EVA-SHUTTLE

The aim of project "EVA-Shuttle" is to open up the first and last mile from the bus station to the house and thus make local public transport more attractive. Autonomously driving mini buses will fill the gap between houses and mobility intersections. Therefore, the project partners set up a fleet testing of electrical, networked and automated electric mini buses and test it under real life conditions on the Test Area Autonomous Driving Baden-Württemberg. Complex, highly automated driving functions with AI methods can thus be tested in a practical way as well as in an observable testing environment.

iBOSS3 – Intelligent Building Blocks for On-orbit-Satellite Servicing and Assembly

Normalisation, standardisation and modularisation can significantly reduce costs in the field of satellite maintenance. Particularly the modularisation shall enable the exchange of damaged or outdated satellite modules in the orbit, instead of replacing them as it was previously the case. The long-term objective of this project is the execution of these works by service satellites.

In the manipulation of cube satellites in the space, satellites have to be exactly positioned in a force-based way. Instead of traditionally relying on teleoperation, iBoss-3 results enable this operation through a neural network. It has learned how to use the sensor values of force/torque sensor systems, in order to join the satellite autonomously and to consider significant framework conditions, which can appear through various configurations, such as already installed, neighbouring satellites.

Facts about the project

  • Technology fields: Robotics & Autonomous Systems, Machine Learning
  • Application market/sectors:  Production, other services, across all industries
  • Application field Industry:  Intelligent Assistance Systems, Robotics, Intelligent Automation
  • Development partners:  TU Berlin, RWTH Aachen University,  RIF Institut für Forschung und Transfer e.V., JOERG KREISEL International Consultant (JKIC)
  • Funding authority:  Federal Ministry for Economic Affairs and Energy
  • Contact person: Arne Rönnau

intelliRISK – Self-confident, autonomous Robots through intelligent risk assessment

intelliRISK aims to enable robots to act more autonomously and flexibly on planetary exploration missions.

The aim of the project intelliRISK is to train robots in such a way that they can assess risks independently and have the ability to consciously weigh up situations and make decisions. The project uses LAURON V, a walking robot developed at the FZI, which is capable of moving safely even on rough terrain.

Facts about the project

  • Technology fields: Image Recognition an Interpretation, Sensors and Communication, Robotics and Autonomous Systems, Machine Learning, Visual Computing
  • Application market/sectors:  across all industries
  • Application field Industry:  Intelligent Sensors, Robotics
  • Development partners:  DLR RM – German Aerospace Center Space Administration
  • Funding authority:  Federal Ministry for Economic Affairs and Energy
  • Contact person: Arne Rönnau

Camera-based driver state monitoring ­

A camera-based system was developed at the FZI in order to gather as thorough an estimate as possible of the state of the driver. This system uses a camera to capture and follow the driver’s face inside the vehicle and evaluates it with regard to his heart rate, blink rate, head position and emotions. The special feature of the approach is that a regular commercial web cam, without additional technology or lighting, can detect the driver’s body, which is a cost-efficient method to use in the vehicle. Heart rate is captured visually with an analysis of pulse-dependent colour changes in the driver’s face. Relevant facial areas are analysed using innovative vision and signal processing algorithms. Blinks, mimic recognition and estimation of the head position are assessed using characteristics in a model of the driver’s face using appropriately trained classifiers.

KarAMeL³ – Karlsruhe Applied Machine Learning Living Lab

The Karlsruhe Applied Machine Learning Living Lab (short for KarAMeL³) transfers knowledge to students in the current research field of machine learning and its practical use.

In the context of final theses, study-accompanying internships, regular workshops, and in non-university competitions, students can gain practical knowledge in the field of machine learning in various application fields. In addition, hands-on workshops for the qualification of future specialists in the current fields of machine learning, such as deep learning, are carried out.

Facts about the project

KolRob – The collaborative, intelligent robot colleague for specialised workers of medium-sized companies

The project KolRob focuses on the problems arising in the human-machine cooperation in industrial assembly. Lacking technologies and concepts, which have been impeding the full automation of assembly processes in the industry, are being identified thereby.

The core research area is, on the one hand, the classification of assembly tasks and the evaluation of robotic systems' feasibility for complex assembly tasks. On the other hand, the possibility of a cooperative assembly by means of concrete application scenarios shall be further developed. 3D workspace monitoring is just as important as the efficient distribution of assembly steps in the process. In cooperation with SMEs and large-scale enterprises, the limits of automation shall be newly defined and innovative, robotic solution approaches shall be transferred from research into industry.

Facts about the project

  • Technology fields: Image Recognition & Interpretation, Human-Machine Interaction & Assistance Systems, Robotics & Autonomous Systems, Virtual and Augmented Reality, Machine Learning
  • Application market/sectors:  Manufacturing Industry/Production, across all industries
  • Application field Industry:  Intelligent Assistance Systems, Robotics, Intelligent Automation
  • Funding authority:  Order of the Baden-Württemberg Stiftung gGmbH
  • Contact person: Arne Rönnau

MikroBo – Microelectronics for the continuous, non-invasive blood pressure measurement in the ear

The aim of the project MikroBO is to develop an electronic microsystem that enables a continuous, non-invasive blood pressure measurement in the ear. Through an active increase of the pressure in an air chamber of the external auditory canal, the blood pressure can be measured. A hearing aid-platform, which is worn behind the ear, serves for controlling and evaluating. For the determination of the absolute blood pressure, innovative algorithms based on machine learning methods are used. Such a system allows a stress-free long-term blood pressure measurement and thus constitutes a substantial progress in the diagnosis and therapy of cardiovascular diseases. The planned miniaturisation allows the use in the ear, so that the blood pressure can be continuously monitored.

 

NeuroReact – Real-time neural planning for reactive industrial robots

In the project NeuroReact, neuromorphic hardware is combined with research results from neurorobotics, in order to realise real-time, reactive planning systems for industrial robots.

Neuromorphic hardware are systems, which are based on the working method of the human brain with networked, artificial neurons. They distinguish themselves through fast, highly parallel and energy-efficient computational processes and thus enable completely new applications and methods for artificial intelligence.

Facts about the project

  • Technology fields: Sensors & Communication, Human-Machine Interaction & Assistance Systems, Robotics & Autonomous Systems, Machine Learning, Visual Computing
  • Application market/sectors:  Manufacturing Industry/Production, Logistics, across all industries
  • Application field Industry:  Intelligent Automation, Robotics, Intelligent Sensors, other
  • Funding authority:  Order of the Baden-Württemberg Stiftung gGmbH
  • Contact person: Arne Rönnau

ROBDEKON – Robot Systems for the Decontamination in Misanthropic Environments

When chemically contaminated areas have to be refurbished or nuclear facilities have to be dismantled, workers are exposed to significant health risks, despite any precautions and protective equipment. ROBDEKON stands for “robot systems for the decontamination in misanthropic environments” and is devoted to the research of autonomous and semi-autonomous robot systems, in order to protect humans from danger zones in the future.

The FZI researches in the framework of the project on how mobile robots can use by means of artificial intelligence automated force-based gripping strategies to recover noxious substances and thus relieve the operator, for example. For the further user support through artificial intelligence, the FZI works on the topics of 3D environmental detection, safe navigation, but especially on intuitive input methods with augmented or virtual reality technologies.

Facts about the project

  • Technology fields: Image Recognition & Interpretation, Sensors and Communication, Human-Machine Interaction and Assistance Systems, Robotics and Autonomous Systems, Virtual and Augmented Reality, Machine Learning, Visual Computing
  • Application market/sectors:  Energy & Environment, Administration & Safety and Security, across all industries
  • Application field Industry:  Intelligent Assistance Systems, Robotics, Intelligent Sensors, other
  • Development partners:  Fraunhofer IOSB, KIT, DFKI, Götting KG, Kraftanlagen Heidelberg GmbH, ICP Ingenieurgesellschaft Prof. Czurda und Partner mbH, KHG Kerntechnische Hilfsdienst GmbH
  • Funding authority:  German Federal Ministry of Education and Research
  • Contact person: Arne Rönnau

Smart Data Accompanying Research

The German Federal Ministry for Economic Affairs and Energy has established an accompanying research for the technology programme "Smart Data – Innovations in Data". The accompanying research supports the funding projects of the technology programme in the efficient implementation of their projects. It networks the individual funding projects with each other and ensures a broad transfer of valid results into the market. In order to achieve these objectives, the accompanying research implements various individual measures:

The accompanying research conducts a systematic, continuous and cross-project monitoring on big data technologies and their economic use in AI applications of the industry, in the fields of energy, healthcare or in mobility projects. The funding projects get also an advisory support through the accompanying research. In this way, an efficient and successful pursuing of the project objectives shall be supported.

The accompanying research initiates, organises and presents working groups on different topics across various fields regarding all funding projects and addressing the development of smart data technologies, such as questions of data protection, data security, acceptance or new business models.

Through the accompanying research, networks are being built, strengthened and supported for the cross-project coordination and the facilitation of rapid and targeted transfer of knowledge. Core elements for this are the website www.smart-data-programm.de as well as the internal cooperation platform. In addition, the accompanying research carries out numerous events, like congresses and professional workshops. Furthermore, they participate in relevant exhibitions and support the networking with associations and other initiatives.

In order to strengthen the macroeconomic impacts of the funding initiative, the accompanying research ensures a widespread communication of the results through numerous measures of public relations and technology transfer. This includes target-group-specific transfer measures, the preparation of publications, like brochures, flyers, fact sheets, newsletters as well as the publication of scientific studies, lectures and articles.

Ultimately, the accompanying research takes measures, in order to ensure the sustainability of the funding priority. For this purpose, concepts are developed as to how the networks established during the project duration between the actors can be maintained and further developed even after the end of the funding phase.

The essential responsibility for the accompanying research lies with the consortium leader FZI Research Center for Information Technology. Considering the networking, the FZI is supported by the German Informatics Society (GI) and considering the technology and result transfer through LoeschHundLiepold Kommunikation (LHLK).

SmartEPark – Smart. Electric. Parking.

In the framework of the project SmartEPark, concepts and methods for an intelligent car park management, cooperating with autonomous vehicles, are researched, in order to enable the effective use of parking spaces and charging infrastructures. The car park management meets the requirement of considering vehicle systems of different degrees of autonomy. The communication between vehicles and the infrastructure enables the development of a common environmental model based on an open map presentation. The concepts are evaluated during the development in the simulation and in real car parks. The functional safety of the system is analysed thereby.

StreamPipes – Self-service Solution for the Manufacturing Industry

StreamPipes offers an integrated self-service solution that makes complex data analyses accessible to end users – from rule-based monitoring and image recognition using artificial intelligence to complex real-time pattern recognition. StreamPipes integrates real-time data sources, algorithms and data sinks in a single application and enables professional users to define data analysis pipelines with a graphical drag-and-drop interface without further development effort. The added value for the end user is the high flexibility, leading to a better data understanding. End users can thus autonomously define, for example, undesirable situations, recognise them at an early stage and derive decisions. At the same time, StreamPipes reduces the effort of providing a fail-safe big data infrastructure for continuous data analyses.

Test Area Autonomous Driving Baden-Württemberg (TAF BW)

The Test Area Autonomous Driving Baden-Württemberg (TAF BW) serves as a real laboratory for research and innovation to test vehicle systems for automated and networked driving in the real-life road traffic. Companies and research institutes can test future-oriented technologies and services for networked and automated driving in everyday traffic, for example automated driving of cars, buses or commercial vehicles such as street cleaning or delivery services. In addition, the regulatory and legal requirements can be developed further. For this purpose, traffic areas of all kinds are equipped with sensors for real-time acquisition of the traffic and its influencing factors, communication hardware are installed as well as highly precise 3D maps are created. Moreover, the data collected can be made available to interested industrial and research partners.