Achtung:

Sie haben Javascript deaktiviert!
Sie haben versucht eine Funktion zu nutzen, die nur mit Javascript möglich ist. Um sämtliche Funktionalitäten unserer Internetseite zu nutzen, aktivieren Sie bitte Javascript in Ihrem Browser.

Info-Icon This content is partly available in English
Show image information

Open Theses

[Translate to English:] Anomalieinjektion in autonomen Systemen am Beispiel eines Miniatur Smart-Factory-Szenarios

[Translate to English:]

Ansprechpartner: Florian Kraus

Der ständige Fortschritt in der Entwicklung neuer Technologien und ihrer Umsetzung birgt viel Potenzial, besonders für Unternehmen. Gerade die Industrie 4.0, in der immer mehr Informationstechnische Systeme und deren Vernetzung im Vordergrund stehen, sorgt hierbei nicht nur für die schnelle Entwicklung, sondern auch für den Versuch der schnellen Umsetzung neuer Technologien und Ideen. Insbesondere Künstliche Intelligenz und Cyber-Physische Systeme stehen hier im Mittelpunkt.

Allerdings gibt es für diese Systeme große Hindernisse, da sie - einmal in einem Unternehmen oder anderweitig implementiert - mit den Bedingungen der realen Welt zurechtkommen müssen und nicht mehr nur mit vorgesehenen Entwicklungsszenarien und Testfällen.

Unerwartete Ereignisse bzw. jene, die von der Norm in einem solchen System abweichen - sogenannte Anomalien - müssen nicht zwangsläufig schädlich sein, können im schlimmsten Fall aber zu fehlerhaftem Verhalten oder sogar einem Aussetzen des Systems führen, wenn es nicht mit diesen Ereignissen umzugehen weiß.

Da Anomalien eine Ausnahme sind und daher eher selten auftreten, stellt sich die Frage, wie testet man ein System gezielt auf seinen Umgang mit unerwarteten Ereignissen?

Die Lösung zu dieser Frage sind gezielt erzeugbare Anomalien, die in ein System bzw. in dessen generierte Daten eingefügt werden können.

Aufgabenbeschreibung:

  • Analyse möglicher Arten von Anomalien anhand des Testsystems des KI4AS-Projekts.
  • Entwurf und Implementierung eines Frameworks zum erzeugen und injizieren von anpassbaren Anomalien.
  • Evaluation auf den Datensätzen des Testsystems.
[Translate to English:] Masterarbeit: Model-Based Product Configuration in Augmented Reality Applications

[Translate to English:]

Contact Sebastian Gottschalk, Enes Yigitbas

Student Eugen Schmidt 

Abstract

Follows...

[Translate to English:] Masterarbeit: Augmented Reality Assisted Robot Programming

[Translate to English:]

Augmented Reality (AR) is a user interface metaphor, which allows for interweaving digital data with physical spaces.AR relies on the concept of overlaying digital data onto the physical world, typically in form of graphical augmentations in real-time. Augmented Reality (AR) techniques are applied in different domains (engineering, entertainment or education & training) and provide a cognitive assistance for various activities (maintenance, programming, configuration etc.). Especially in the context of robot programming, AR provides several advantages in the programming process. The operator gets instant real-time, visual feedback of a simulated process in relation to the real object, resulting in reduced programming time and increased quality of the resulting robot program. Main goal of this thesis is to design and implement an augmented reality assisted robot programming environment.The robot programming environment shall focus on the DobotMagician system which can be represented as a 3D Model and projected via AR techniques in the working environment to support the programming task of an operator.

Task:

  • Literature research on the topics AR and Robot Programming
  • Conception and Design of an AR Assisted Robot Programming Environment
  • Prototypical Implementation of an AR Assisted Programming Environment for the Dobot Magician System
  • Experiment-based evaluation of the implemented approach

Further details regarding the topic and tasks can be found in the attached PDF file.

Contact

Enes Yigitbas

Download

[Translate to English:] Masterarbeit: Development of a virtual reality interface for human-robot interaction

[Translate to English:]

Motivation

Cognitive assistance systems offer a variety of possible use cases especially in the context of learning environments in artificial reality (AR/VR) scenarios. In our project it‘s OWL AWARE , we are researching new application scenarios, development methods, and new forms of interaction to support the development of novel assistance systems.
In this subject area, we are especially interested in the topic of human robot collaboration. With the means of virtual reality interfaces it is possible to control robots remotely, i.e., to teleoperate them. The teleoperation of robots with virtual reality brings lots of advantages regarding safety, supervision from distance etc.
The main goal of this thesis is to design and implement a virtual reality environment which supports remote control of the Dobot Magician system (https://www.dobot.de).

Description of the Task

  • Systematic literature survey on VR interfaces for remote control
  • Prototypical development of a VR application for remote control of Dobot Arms
  • Usability evaluation of VR application in terms of efficiency, effectiveness, and user satisfaction

Contact

Enes Yigitbas

Download

Ausschreibung

[Translate to English:] Masterarbeit: Systematic Literature Review on Tolerance in Model-driven Engineering

[Translate to English:]

Contact: Nils Weidmann

Motivation:

Tolerating inconsistencies – or flexibility in MDE in a broader sense – is a research topic of growing interest. Domain experts shouldn’t be forced to maintain perfect consistency between multiple models at each point of time for several reasons:

  • There might be different equally good solutions
  • There might be no consistent solution at all
  • Restoring consistency involves unexpected changes

Even though the topic is frequently discussed for many years, there is no systematic overview of the proposed literature up to now.

Tasks:

  • Conduct a systematic literature review on tolerance in model-driven engineering according to a standard research method in software engineering
  • Develop a tool chain for collecting papers from a research database and filtering them according to given criteria
  • Compose a suite of examples to demonstrate fault-tolerant approaches and prototypically implement some of these examples in an Eclipse-based MDE tool

Preconditions: Successful participation in at least one of the courses:

  • Bachelor Lecture: Model-Based Software Development
  • Master Lecture: Model-Driven Software Development
  • Master Lecture: Fundamentals of Model-Driven Engineering
  • Seminar: Advanced Model-Based Techniques
  • Seminar: Maintaining Consistency in Model-Driven Engineering
  • Project Group: VICToRy
[Translate to English:] Masterarbeit: Step-wise Visualization of Model Transformation Processes

[Translate to English:]

Contact: Nils Weidmann

Motivation:

Model transformations are an important aspect when maintaining consistency in the field of Model-Driven Engineering. Classic MDE tools treat model transformation as a background task without user interaction or an appropriate visualization of the transformation results. However, these features would be beneficial for the users of the tool to understand both the process and the output models – and as a result – consider the MDE tool as trustworthy software. The project group “VICToRy” started developing a debugger for model transformations in winter 2018/19 and summer 2019.

Tasks:

  • Create an overview of breakpoint types used in MDE and other fields of software engineering, such as object-oriented or functional programming
  • Create a detailed concept for breakpoints for model transformations using software engineering methods
  • Enhance the VICToRy debugger by breakpoint functionalities
  • Evaluate the approach by conducting a case study with students of an MDE-related lecture

Preconditions: Successful participation in at least one of the courses:

  • Bachelor Lecture: Model-Based Software Development
  • Master Lecture: Model-Driven Software Development
  • Master Lecture: Fundamentals of Model-Driven Engineering
  • Seminar: Advanced Model-Based Techniques
  • Seminar: Maintaining Consistency in Model-Driven Engineering
  • Project Group: VICToRy
[Translate to English:] Masterarbeit: Integration of Domain Constraints into a Tool for Bidirectional Model Transformations

[Translate to English:]

Contact: Nils Weidmann

Motivation:

Domain constraints that go beyond restrictions already encoded into meta-models are essential for modelling software systems of practical relevance. Rule-based approaches to MDE often require the user to indirectly specify these domain constraints as application conditions that are attached to single rules instead of directly specifying them for the whole system. This involves several problems: it is particularly difficult to validate the completeness of application conditions, especially when new rules are added to the system. An approach for directly integrating domain constraints in rule-based model transformations was already presented on a theoretical basis but lacks implementation for more than simple consistency checks.

Tasks:

  • Integration of domain constraints for different consistency management operations, including forward and backward transformation as well as consistency checking by correspondence creation
  • Enhancing the eMoflon Specification Language (EMSL) to specify domain constraints and create an appropriate visualization for the user
  • Evaluate the approach regarding correctness, expressiveness (in comparison to other approaches) and scalability

Preconditions: Successful participation in at least one of the courses:

  • Bachelor Lecture: Model-Based Software Development
  • Master Lecture: Model-Driven Software Development
  • Master Lecture: Fundamentals of Model-Driven Engineering
  • Seminar: Advanced Model-Based Techniques
  • Seminar: Maintaining Consistency in Model-Driven Engineering
  • Project Group: VICToRy
[Translate to English:] Masterarbeit: Consistent transformation and synchronization of UML-B and Event-B models

[Translate to English:]

Contact: Nils Weidmann

Motivation:

Safety-critical systems have to be verifiable, meaning that the correctness of the software system can be formally proven. In the area of software modelling – you require models to have an unambiguous formal semantics in order to conduct verification steps. However, ambiguous visual languages like SysML (Systems Modeling Language) are well-accepted in industry and therefore many engineers are far more experienced with languages that do not fulfill the requirements to be verifiable.

Adding knowledge about the application domain, it is possible to consistently transform models in SysML notation into Event-B to formally verify the created system models. In case the Event-B models must be adapted due to the verification results, a backward transformation to SysML should also be possible. To keep the transformation flexible and to align the two directions of transformation, a bidirectional rule-based approach should be used.                                                       

Tasks:

  • Specify transformation rules for SysML to Event-B
  • Implement a bidirectional transformation between the two language within an MDE tool
  • Create a test suite of models to validate the approach

Preconditions:

Successful participation in at least one of the courses:

  • Bachelor Lecture: Model-Based Software Development
  • Master Lecture: Model-Driven Software Development
  • Master Lecture: Fundamentals of Model-Driven Engineering
  • Seminar: Advanced Model-Based Techniques
  • Seminar: Maintaining Consistency in Model-Driven Engineering
  • Project Group: VICToRy
[Translate to English:] Masterarbeit: Rule-Based Product Configuration in Client-Server Systems

[Translate to English:]

Contact: Nils Weidmann

Motivation:

Rule engines are used for validation tasks for product configurations in various branches of producing industry. In contrast to directly programmed validations, rule-based approaches have the advantage of being flexible for additional or changed requirements that arise while working with a software system. In a distributed scenario, an additional dimension of complexity comes into play: Where does the validation task take place? Conceptually, it’s the easiest way to run all validations on a server, but this usually causes performance problems. So a more efficient solution would be to distribute validations among server and clients, depending on the knowledge and data access these systems have.

Tasks:

  • Develop an architecture for rule-based validation in distributed systems
  • Compare multiple rule engines and choose the most suitable one for client and server side
  • Implement a prototype for rule-based validations on client and server

Preconditions:

Successful participation in at least one of the courses:

  • Bachelor Lecture: Model-Based Software Development
  • Master Lecture: Model-Driven Software Development
  • Master Lecture: Fundamentals of Model-Driven Engineering
  • Seminar: Advanced Model-Based Techniques
  • Seminar: Maintaining Consistency in Model-Driven Engineering
  • Project Group: VICToRy
[Translate to English:] Bachelorarbeit:

[Translate to English:]

Motivation

Wissenssysteme unterstützen Entscheidungsträger nicht nur bei der strukturierten Ablage von existierendem Wissen, sondern ebenfalls bei der Generierung von neuem Wissen aus vorhandenem Wissen. Insbesondere der erste Anwendungsfall setzt voraus, dass Entscheidungsträger ihr Wissen gemäß der genutzten Technologie (wie beispielsweise dem Resource Description Framework RDF [1]) in das Wissenssystem eintragen können. Dazu können Entscheidungsträger sowohl die entsprechenden Beschreibungssprachen als auch Editoren (wie bspw. den Fluent Editor [2] oder Protégé [4] für RDF) nutzen. Diese Ansätze sind jedoch darauf ausgelegt, Wissen unabhängig von der konkreten Domäne des Entscheidungsträgers zu erfassen, was die Komplexität der Sprachen bzw. Tools erhöht und die Benutzerfreundlichkeit meist einschränkt.

 

In der Regel ist es möglich, die Datentypen des in einem Wissenssystems abzulegenden Wissens durch eine Ontologie zu beschreiben (für RDF siehe Web Ontology Language OWL [4]). Im Rahmen der Bachelorarbeit soll evaluiert werden, ob unter Zuhilfenahme einer Ontologie automatisch ein domänenspezifisches Wissenssystem inklusive grafischer Benutzer-oberfläche generiert werden kann, mit dem Entscheidungsträger bei der Wissenserfassung produktiver arbeiten können.

 

Aufgabenbeschreibung

Ziel der Bachelorarbeit ist es, zu evaluieren, ob die Generierung einer Web Applikation (Front- und Backend) zur Erfassung und Persistenz von durch eine Ontologie beschriebenem Wissen zum einen technisch möglich ist und zum anderen die Produktivität bzw. das Anwendererlebnis von Entscheidungsträgern und Domänenexperten bei der Wissenserfassung verbessert. Dazu müssen insbesondere folgende Aufgaben bearbeitet werden:

•    Grundlegende Einarbeitung in Wissenssysteme, Ontologien, Usability Metriken [5], etc.
•    Zusammentragen von verwandten Arbeiten (bspw. [6-9]) und Erhebung von Anforderungen basierend darauf
•    Entwicklung eines Konzepts zur Generierung von Wissenssystemen (Front- und Backend) basierend auf einer Ontologie
•    Prototypische Implementierung & Evaluation des Ansatzes gegenüber den zuvor genannten „General-Purpose-Editoren“

 

Referenzen

[1]: www.w3.org/RDF/

[2]: www.cognitum.eu/semantics/FluentEditor/

[3]: protege.stanford.edu

[4]: www.w3.org/OWL/

[5]: www.researchgate.net/publication/220635983_Usability_measurement_and_metrics_A_consolidated_model

[6]: www.researchgate.net/publication/221565287_Ontology-based_Editor_for_Metadata_Documents

[7]: oro.open.ac.uk/29124/

[8]: link.springer.com/chapter/10.1007/978-3-642-28062-7_3

[9]: ceur-ws.org/Vol-1268/paper2.pdf

The University for the Information Society