Activity Recognition is a growing field using ambient or wearable sensors and concepts of machine learning interpreting everyday behaviour in interactive systems. It is about detecting human activities (mostly activities related to physical movements) using miniature sensors, for example motion and posture sensors (accelerometers), placed on the body or in everyday environments.

Activity recognition can be mostly found in two paradigms of human-computer interaction (HCI) and ubiquitous computing: First the concepts of seamless interaction, in which activity recognition are used to support the interaction with technical systems of any kind (computing device, robot, vehicle, aircraft) in an unobtrusive way. Activity-sensitive systems can support users in archiving their goals by reducing the need of explicit inputs. Second, is about to use activity recognition with the purpose to present the results to the users of the devices with the future goal of supporting self-reflection. To use activity recognition for self-tracking and further for self-reflection is already common in the field of sport life-style products where e.g. Nike Fuelband, Fitbit or Jawbone UP and apps on smartphones been used. For example, there are research projects where sleeping behaviour is logged, smoking behaviour detected or working steps in the laboratory documented to become more aware of this activities with different goals of changing them in a good way or analysing the work process.

In the PhD project, the technology of activity recognition is analysed on a conceptual level as well as discussed in the context of human-computer interaction design. With focus on the technologies, general thoughts on sensor-systems and concepts of machine learning for detection (together called as machine interpretation) are built further context. For that, the relation of humans and technical systems is conceptualised based on philosophical theories (embodiment and media philosophy). The technical aspects and the conceptual understanding are brought together for a mutual interactive design theory. The central goal is to model the effects (based on the technical mediation) on human behaviour and perception while interacting in and with a technical sensing and interpretation systems. The follow up question is then, how to design the interfaces between users and the technology which senses and interprets “autonomous”, to make a trustful and insightful interaction as well as self-perception possible.

On the level of system design and empirical evaluation an interactive system is implemented which makes it possible to visualize the daily behaviours of users in various ways. This is linked to different applications in the fields of quantified self and health care. The technical system design and conceptual work is and will be realized in different projects on activity recognition: for example in psychology trails [1], as well for diabetes self-management and therapy for chronic pain patients.  [1] Dietrich, M., Berlin, E., & Laerhoven, K. v. (2015). Assessing Activity Recognition Feedback in Long-term Psychology Trials. In Proceedings of the International Conference on Mobile and Ubiquitous Multimedia (MUM), ACM, Linz, Austria.