Project Details
Projekt
DEUS: Debugging for End Users in Smart Environments
Subject Area
Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
Term
since 2025
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 562955803
The task of debugging smart home setups is spinning out of control. As smart appliances are rapidly entering the home, with them comes the task for inhabitants to continuously debug them: first until all automations work as intended, then again as components are added, fail, or users’ needs change. Debugging becomes an ongoing housekeeping chore, frequent enough to annoy, yet rare enough to forget the intricacies of one’s own setup. To make matters worse, smart home users are not trained programmers, and they live inside the environment they are debugging, making testing nearly impossible, and the effects of mistakes (like bedroom lights turning on at 3am) very real. The problem expands beyond the smart home to public environments like hotel rooms with unintelligible automations, and the workplace where building automations regularly ignore users’ individual needs and provide little agency. In all, scalable deployment of smart technology across these environments, to improve quality of life and reduce emissions, is severely limited by being almost impossible for end users to debug today. The DEUS project tackles this challenge by (1) unveiling users’ mental models of smart home automations, (2) documenting their debugging strategies, and (3) proposing new interaction techniques and concepts for digital tools to better support these tasks. We then (4) design and implement research prototypes to (5) scientifically validate our findings, tools, and techniques in user studies. We structure our research using three overarching research methods, M1–M3. First (M1), across the entire project duration we progress through seven increasing levels (L1–L7) of automation complexity, starting in WP2 with simple, natural state–state (L1) and event–action (L2) mappings. In each following WP, we apply a similar research approach, but to a significantly more involved problem space. Because of this, WP2 (L1+L2), WP3 (L3–L5), and WP4 (L6+L7) are similar in structure, but qualitatively different in scope. Within each of those WP, we first take a black-box approach to identify users’ mental models of the smart home and its faults at that level of complexity, then later apply a white-box approach to understand their debugging strategies (M2). Finally, each of these investigations is structured along an error space (M3) that classifies the fault situations users may encounter.
DFG Programme
Research Grants
