Student Projects

Are you interested in writing your Bachelors/Masters/Diploma thesis here at HCI? Please, have a look at the following list of possible topics or fields (plus a responsible contact) or browse through the theses in progress or finished work. Feel free to have a look at our research projects, too.

LaTeX Template for Bachelor's or Master's Theses

We offer a LaTeX template to support you writing your bachelor's, master's or diploma thesis. The template can equally be used for German and English theses and comprises common elements, such as a title page and the confirmation statement, as well as the usually used styles and formats. You can download, clone and fork the template from its GitHub repository TUD-HCI-Thesis.

Furthermore, you will find information on creating accessible documents on the website of the Working Group Services Disability and Studies (AG SBS). Especially for those who want to create accessible PDF documents using Word, PowerPoint and InDesign, there are instructions available directly within the download section of the AG SBS.

Ideas for New Theses

Gesture recognition for controlling common Office applications for blind and severely visually impaired people

Contact person: Bela Usabaev

Keywords: Human-Computer-Interaction, Haptification

In order to navigate the visual desktop interface, blind and severely visually impaired people use screen readers such as NVDA to output the displayed content using speech. A Braille display or a standard keyboard is usually used for input.
On tablets and cell phones, blind users control applications (apps) with gestures such as “tap” and “swipe”, which correspond to the “Enter”, “Tab” or “Arrow” keys, and receive feedback by voice [or Braille via a Braille display].

In the field of education, tactile graphics are often used to convey graphically coded complex facts, such as maps or diagrams, through the sense of touch and explanatory voice output. With the help of computer vision, it is possible to implement gesture control on a tactile graphic via a camera image.

In the course of developing a voice assistant for tactile graphics as part of the project TactonomDuo, we want to investigate which gestures are useful for common Office programs such as Excel, Calendar and Explorer, which are often used in everyday professional life, and develop and evaluate a gesture vocabulary for controlling these applications.

Link to project web site: TactonomDuo

The following topics are of interest in this context:

• Gesture recognition from a camera image with the Raspberry Pi

Accessible teaching: checklist-based self-assessment on the accessibility of course materials, sensitization of lecturers

Contact Person: Emma Etzold

Learning management systems (LMS) such as OPAL allow teachers to make their course materials available online. They can choose from a variety of options, from uploading files that students can download to self-programmed website content embedded in the LMS. Teachers are responsible for how accessible their materials are. This means that they decide whether and to what extent, for example, alternative texts for images or video transcriptions exist. Depending on the scope and type of materials, different accessibility criteria are relevant.

In Germany, it is a legal requirement that public institutions - including universities such as the TUD - must make their digital content accessible. This means that they must consider the criteria of the Barrierefreie-Informationstechnik-Verordnung (BITV). These criteria are usually difficult to understand and are primarily aimed at experts - web developers or accessibility specialists. However, many university teaching staff are not among these specialists and are either unfamiliar with the specific guidelines or only partially. Nevertheless, many of them are endeavoring to increase the accessibility of their course content and make learning accessible to all students. To support them in this, the chair has developed a concept and a prototype for a checklist-based self-assessment of a course's accessibility. The first step is to create an "accessibility statement" that informs students about how accessible or inaccessible course materials are. This should also create an opportunity to report existing barriers directly to the responsible person and to remove them as quickly as possible or to obtain alternative access to the information. We also want to use this checklist to sensitize and motivate lectures to accessibility. That is why, in addition to the checklist, there is also a handbook that teachers can use to learn about individual criteria in an easily understandable way and learn how to implement them in an LMS such as OPAL.

The following topics and questions are of special interest for theses on this topic:

• What checklists already exist for specific file formats?
• How easy are such checklists to understand for non-specialists?
• How can BITV criteria be implemented in LMS (OPAL, Ilias, moodle) to create accessible online courses?

Supporting People on the Autism Spectrum at Work by Improving Work Organization and Reducing Stimuli 

Contact:   

Mental illness and developmental disorders are often related to a reduced ability to work and adapt, which often leads to reduced employability. People with autism are much more likely to be affected by unemployment or to be employed in jobs with low qualification requirements even though they have a good vocational or school education compared to people without autism. Challenging working conditions such as time pressure, information overload, high communicative demands, social skills, sensory overstimulation, and the demand for a high degree of flexibility are the major reasons for the low employment rate of people with autism in the primary labor market. Until now, only a few technical aids have been available to profitably support people with autism in their everyday work.   

The field of research comprises the development of technical assistance systems to support task and time management on the one hand, and tools to support stimulus reduction on the other. The aim is to develop an adaptive noise cancellation method with ANC headphones that, for example, detects the context of a specific conversation situation and adaptively filters out irrelevant signals. Concepts for the use of data glasses, e. g. for person identification or emotion recognition, will be developed as well. In addition, haptic, natural interaction modalities using tangibles are to be used to improve task and schedule planning and thus enable structured work with minimal disruptions.  

Theses in the research field should support at least one of the three problem areas in particular:   

• Stimulus Reduction: Reduction of stimuli to avoid overstimulation  

• Communication: Support of verbal and textual communication  

• Task Management: Support of structuring and prioritizing in task and time management  

Depending on the specific scientific question, the focus of this work can be on technical components and their possible applications as well as on concepts for supporting people on the autism spectrum at work.   

Project website with further information: https://tu-dresden.de/ing/informatik/ai/mci/forschung/forschungsgebiete/autark?set_language=en  

Multimodal presentation and exploration of time-dependent real-time data by blind people

Contact Person: Emma Etzold

Keywords: Human-Computer-Interaction, multimodal data presentation, Haptification, Sonification

The visualisation of dynamic data is a standard method to analyse and derive information from them quickly. When monitoring time-dependent real-time data streams (e.g., machine monitoring in a production context), visualisations (e.g., dashboards) are often used to quickly overview the data volume and temporal progress. Most data representations are purely visual and use visual abilities to convey information from data quickly. However, a purely visual data representation excludes blind people from professional data monitoring.

In this research area, multimodality is used to enable greater inclusion of blind people in data monitoring. The aim is to develop a multimodal, non-visual method for exploring real-time data streams using off-the-shelf mobile devices and wearables, enabling blind people to analyse them independently.

In this context, the following research questions are interesting:

• Which (multimodal) approaches of data presentation are equally suitable for blind and sighted people? (Information/data visualisation)
• Which non-visual output media can be usefully combined with each other? (Perception and interpretation of output media)
• How can the user’s attention be focused on certain events (e.g., exceeding thresholds)? (human-activity-recognition)
• Which input modalities can be usefully combined with each other? (interaction)

Conversational Agents in Mental Health Care

Contact Person: Julian Striegl

With rising numbers of people with mental health problems worldwide and an already struggling healthcare and metal health system, new solutions have to be found to improve the mental wellbeing of people from all age groups. In the past, computerized therapy approaches – such as application-based and internet-based cognitive behavioral therapy (CBT) – have already been investigated and have shown to be effective for several different forms of mental disorders, such as depression, anxiety disorder or obsessive compulsive disorder.

In the last decade the research interest in chatbot-based CBT has grown, with voice assistant-based CBT (VA-based CBT) being a possible next step towards offering scalable, low threshold, easy to use and accessible solutions for mental health support. How can we create conversational agents (such as chatbots and voice assistants) that are able to support the mental wellbeing of users through methods based in psychotherapy, positive psychology and mental health coaching?

Subject areas in this field of research include: artificial intelligence and machine learning, natural language processing, voice user interface design and interaction design combined with elements of clinical psychology and psychotherapy. Furthermore, several research questions can be derived from the central topic:

• How can we design suitable voice user interfaces that take the diverse needs of people with different communication needs and capabilities into account in order to provide accessible, low threshold therapy support?
• How does the usability and acceptance of VA-based CBT compare to chatbot-based CBT?
• Which methods from fields such as CBT and positive psychology are suitable to be presented/administered via conversational agents?
• How can we support salutogenesis through chatbots and voice assistants?
• How does an emotion sensitive dialog management affect the acceptance of and adherence to VA-based CBT?
• How does the effectiveness of VA-based CBT compare to alternative application-based and internet-based CBT approaches?

Accessible Maps for Buildings

Keywords: Accessibility, Indoor Maps

The research field Accessible Maps for Buildings addresses different problems in the field of ​​indoor maps. Therefore, problem statements such as the acquisition of building data and their use are considered. On the one hand, the resulting research field ranges from the use of computer vision to transforming escape plans into digital building plans, to SLAM processes which are used to improve maps. On the other hand, it discusses the area of ​​user-centered design to generate maps for specific user groups or develop routing applications.

Cartography of Buildings

Contact Persons: Jan Schmalfuß-Schwarz Julian Striegl

The research area of ​​the cartography of buildings deals with the generation, modelling, improvement and updating of indoor map data. In this regard, card formats are evaluated and developed which, in addition to the purely geometric representation of the building, also store information about accessibility and discuss approaches for capturing time-critical (dynamic) barriers in indoor maps. Moreover, the focus lays on solutions that make it possible to extract indoor maps from various output formats, such as architectural building plans or escape plans, using image processing and deep learning methods. Furthermore, the different solutions have to be examined in regards to their accuracy and optimized if necessary. Therefore, approaches from robotics as well as from the crowdsourcing area are used.

Usage of Map Data

Contact Persons: Julian Striegl Christin Engel

In addition to the creation of visual building maps, the usage of such maps is moreover a diverse topic. Both the generation of physical maps and their mobile use are examined using various modalities. On the one hand, it focuses on the generation of user-centered maps that address the specific needs of people with disabilities. This includes the generation of barrier-free maps (for example in the form of tactile or physical maps for haptic exploration). On the other hand, the goal of the research field is to explore mobile interaction techniques with digital maps to ensure user-friendliness for a variety of use cases and different contexts as well as user groups. This includes – among other topics – the design and creation of voice user interfaces and the accessible representation of (and interaction with) digital map data using screen readers. In this context, the use of location-based services in connection with indoor maps is also examined.

Diversity-Sensitive Human-Computer Interaction and AI

Contact Person: Claudia Loitsch

Keywords: human-computer interaction, user modeling, adaptive systems, human-cantered design, diversity-sensitive design, user experience, user research, AI.

The penetration of digitalization into all areas of life and even all age groups has led to the situation that we now perform almost all tasks in some form of computer-assisted. This means that people, with their individual experiences, expectations, abilities and needs, play an even more important role in the development of computer-based solutions and user interfaces. The overall user experience and digital participation have an increasing economic and social relevance.

This topic investigates how diversity, or the variety of people, can be taken into account in the human-cantered design of computer-based systems and AI.

The idea is that computer-based systems and applications should adapt to our individual needs, abilities, and mental models to help us perform everyday tasks as efficiently as possible and, increasingly, in enjoyable or fun ways. This requires that digital applications and user interfaces become more intelligent and can adapt to our individual characteristics, abilities or life situations. It is also increasingly important that key technologies such as artificial intelligence take diversity into account, in particular that training data is fair and balanced, and does not promote discrimination and exclusion. 

Topics will be developed with you depending on your interests. A wide range of questions can be considered in this topic area, such as:

• How do different expressions of personality traits (e.g., Big Five/OCEAN Model) affect digital design solutions?
• Which perceptual or behavioural psychological aspects, design guidelines, design decisions can be applied for diverse target groups (e.g. elderly people, people with migration background, people with cognitive impairments, etc.)?
• How can diversity be achieved and detected in training data for machine learning and AI?
• Which user models address diversity of groups or individuals that lead to a better user experience in adaptable or adaptive applications and user interfaces?

Social Assistance Robots ― Usability and Software Variability

Contact Person: David Gollasch

Keywords: Human-Robot Interaction, Robotic Cooperation, Emotion Recognition and Modelling, Software Variability, Software Ecosystems, Voice Interaction, Gesture Control, Smart Home, Adaptivity, Accessibility, Assistive Systems

Assistive technology improves or enables the participation in society as well as coping with everyday life for many people. Especially the trend of an aging society and the desire for greater inclusion leads to more significance for the field of developing powerful assistive technology. A subset of such technology is made up by social assistance robots (SARs) which are developed for various different tasks and purposes. One key issue in this regard is how to efficiently develop software for those assistive robots such that our target group of people with special needs are able to interact well with them. Thus, we need to design feasible and reliable development methodologies that lead to an appropriate human-robot interaction.

Furthermore, there arises a variety of interesting research questions. Here are some examples:

• How should diversity-sensitive voice interaction look like (related fields: voice user interfaces, conversational user interfaces, artificial intelligence)?
• How to implement assistance robot activities that dynamically match specific requirements (related fields: adaptivity, UI software variability, artificial intelligence)?
• What input modalities should be supported by an assistance robot (related fields: gesture controle, emotion recognition and modelling)?
• How to embed the assistance robot into its environment (related fields: smart home integration, context-sensitivity, ambient assisted living, ADL recognition)?

Improving Presence Lecturing for Blind and Visually Impaired People 

Contact Person: Jens Voegler 

During lectures, exercises and seminars, interactive whiteboards and slide shows are used quite regularely. By using latest Web technologies, the accessibility of those media for blind and partially sighted people can be improved.

Physical Information Visualisation

Contact Person: Meinhardt Branig

Physical Information Visualisation for Big Data Exploration for Blind and Visually Impaired People.

Completed Theses

As soon as theses are listed within the TUD Research Information System, they will be listed below.