"Hello, zoom in!" - Navigation in virtual space
Adrian Stock, a student at the Baden-Wuerttemberg Cooperative State University, wrote his bachelor's thesis at the Department of Intelligent and Distributed Systems. His thesis focused on voice-controlled navigation in augmented reality. Adrian developed and evaluated a voice assistant for IslandViz.
Developing a voice assistant
Virtual and augmented reality (AR) make it possible to visualise complex software architectures in three-dimensional space. This allows users to navigate between the different elements of the software to explore the different components and their interdependencies. The usual navigation in AR is gesture driven. This means that certain hand movements perform certain actions. However, the number of possible actions is limited by the small number of gestures available. To provide more interaction options when exploring IslandViz, Adrian independently developed a voice assistant that can extend or replace gesture-based navigation. Voice commands are interpreted and executed by a Natural Language Understanding (NLU) service. In the future, the voice assistant can be used to perform complex searches to enhance and simplify navigation.
Gesture versus voice control
Adrian conducted a user study to compare the usability of gesture and voice control within IslandViz in AR. 16 test subjects evaluated both the gesture control and the newly developed voice assistant. The study consisted of two main parts. The first part tested the effectiveness of the navigation systems by measuring the time taken to complete three navigation tasks. The second part of the study investigated the subjective satisfaction of the users by means of questionnaires.
Adrian's study found that the usability of gesture control and voice control were not rated differently. Both navigation systems were rated as user-friendly. Satisfaction with both control systems was also very similar, so there was no clear winner. However, the subjects needed more time to complete the tasks with the voice assistant. This could be due to the processing time of the voice commands. As the time taken to complete a task has a significant impact on user satisfaction, Adrian expects that improved processing of voice commands and the associated immediate feedback to the user will increase user satisfaction. There is still a lot of potential in the field of voice control, as Adrian recognises.
"I will miss my time at DLR"
Adrian particularly enjoyed working with his colleagues during his time at DLR. "The working atmosphere was great and I always enjoyed going to work," he says. One of the reasons for this was that Adrian felt like a full member of the team right from the start. He was allowed to attend the weekly department meetings and learned a lot about the department, his colleagues and the many projects throughout Germany. When it came to assigning the topic for his bachelor's thesis and the tasks for the semester, a lot of emphasis was placed on what Adrian was particularly interested in. He enjoyed the topicality of speech processing in relation to AR, but also learnt a lot about software engineering and general teamwork. Adrian now looks back on the past few years with a smile. "I will miss my time at DLR. I have learnt a lot," he says.
Adrian has a lot of plans after his successful graduation. While his dual study programme was very practical, he now wants to get a bit more theoretical. He will be doing his Masters in Machine Learning at the University of Tübingen.