A Player/Stage System for Context-Aware Intelligent Environments Matthias Kranz 1 Radu Bogdan Rusu 2 Alexis Maldonado 2 Michael Beetz 2 Albrecht Schmidt 1 1 {matthias,albrecht}@hcilab.org 2 {rusu, maldonad,beetz}@cs.tum.edu 1 Research Group Embedded Interaction, University of Munich Amalienstrasse 17, 80333 Munich, Germany 2 Technische Universität München (TUM) Boltzmannstrasse 3, 85748 Garching b. München, Germany Abstract. The effective development and deployment of complex and hetero- geneous ubiquitous computing applications is hindered by the lack of a compre- hensive middleware infrastructure: interfaces to sensors are company specific and sometimes even product specific. Typically, these interfaces also do not sustain the development of robust systems that make use of sensor data fusion. In this paper, we propose the use of Player/Stage, a middleware commonly used as a de facto standard by the robotics community, as the backbone of a heteroge- neous ubiquitous system. Player/Stage offers many features needed in ubicomp, mostly because dealing with uncertainty and many different sensor and actuator systems has been a long term problem in robotics as well. We emphasize they key features of the Player/Stage project, and show how ubi- comp devices can be integrated into the system, as well as how existing devices can be used. Additionally, we present our sensor-enabled AwareKitchen environ- ment which makes use of automatic data analysis algorithms integrated as drivers in the Player/Stage platform, of which we are active developers. 1 Introduction Intelligent sensor-equipped environments can be much more helpful if they are capable of recognizing the actions and activities of their users, and inferring their intentions. Understanding human activities and characterizing them into expressive and detailed activity models is one of the key issues of today’s current pervasive computing systems. Integrated ubiquitous computing environments are still rare to find. By integrated we mean that the ubiquitous technology is seamlessly interwoven within a real-world setting and not put into an artificial laboratory or a single dedicated ubiquitous com- puting room. Examples of highly integrated sensor-enriched environments are e.g. the Georgia Tech Aware Home [1] or MIT’s PlaceLab [2]. Those research facilities enable researchers to develop, build and test context-aware applications in a real-world setting. Building and maintaining these environments is expensive, in terms of time and money, and thus, not many researchers are able to work in them. While an invaluable source for researchers, systems and software are often not publicly available, e.g. Mites as novel sensor platform currently cannot be used by other researchers to reproduce research results. Annotated sensor histories are one solution to share data and context information and allow for algorithms and applications development without the need for the original infrastructure. Again, those annotated histories are costly to produce and much effort for labelling is required.