Towards Ambient Intelligence:
Methods for Cooperating Ensembles in Ubiquitous Environments

KI 2007 Workshop

September 10, 2007, Osnabrück, Germany



"Ambient Intelligence" aims at creating everyday environments that assist humans in their daily living activities — at home, at work, while traveling, while learning In recent years, smart environments — smart homes, smart offices, intelligent classrooms etc. — have become an important research area. The central questions investigated are: (a) how can an environment discover the current needs and objectives of the users, and (b) how should the environment employ its actuators in order to provide an appropriate reaction? Research results show that smart environments — infrastructures that are able to proactively assist a user — are indeed possible.

However, today the hardware and the software-infrastructure of such environments is the result of a careful and deliberate design process. Setting up a smart environment and orchestrating its sensing and actuating components is a substantial and expensive endeavor. This may be conceivable for research prototypes and high-end working environments — but surely not for the everyday home.

On the other hand, the ubiquity of computing is increasing without restraint — embedded intelligence will prospectively pervade all kinds of everyday artifacts. This raises an interesting question: is it possible to let the intelligence of an environment emerge from the interaction between the individual smart artifacts that coincidentally reside in this environment? Is it possible to devise strategies that allow such a set of artifacts to form a coherently acting ensemble, without human configuration and supervision?

A specific challenge here is the heterogeneity of knowledge available in the different artifacts: built by different vendors with different objectives, it will be a major task to unify the local knowledge provided by the artifacts. How can a coherent world model be synthesized, or sufficient mutual compatibility of local models be ensured, such that the ensemble is effectively able to base its strategies on a unified knowledge of the user, her needs, the ensemble's capabilities, and the current situation? This is even more challenging, when different artifacts employ different reasoning strategies — such as reactive vs. deliberative approaches.

Answering these questions is of fundamental significance for the future research directions on smart environments and ambient intelligence.


Within the application area of smart environments, the focus of this workshop is the discussion of possible solutions for emergent intelligence in an ensemble of everyday artifacts. Because of their similarities with respect to distributed processing, self-organization, learning, fault tollerance, etc., such ensembles should be able to draw significant inspiration from research on distributed artificial intelligence and multi-agent systems.

Two aspects here seem of central importance: emerging communication languages and cooperative problem solving. The first aspect is concerned with enabling the communication of intelligent entities that have no common a-priori language. Such entities may use different notions for the same concepts (e.g., situations), as well as having only partially overlapping world models. It might be of further interest to explore cases in which different entities communicate even though they do not have any explicit intention to do so.

The second aspect, cooperative problem solving, is devoted to the requirement that in the absence of any central master controller or planning entity, the devices have to analyze the users' intentions as well as the construction of adequate strategies and actions in an entirely distributed form. This distributed problem solving has, among other things, to be able to cope with uncertain, incomplete, and even contradictory knowledge on user's requirements, as well as the possibility of failing devices.

Within these major areas, we solicit position papers discussing relevant issues such as (but definitely not limited to):

All position papers should have a clear relation to the topic of ambient intelligence and smart environments: systems that aim to assist the human user in the real world. We specifically welcome investigations based on case studies and real-world experiments.


The workshop is planned as a full-day workshop with ample of time for talks, discussions, and practical presentations.

It is planned to prepare a workshop report, summarizing the results and the ideas / recommendations jointly developed in the discussions.

Program Committee

Clemens Beckstein, Jena University
Gerald Bieber, Fraunhofer Institute for Computer Graphics Rostock
Andreas Butz, LMU Munich
Michael Hellenschmidt, Fraunhofer Institute for Computer Graphics Darmstadt
Giacomo Indiveri, ETH Zürich
Christopher Lueg, University of Tasmania
Tiziana Margaria, Potsdam University
Rico Möckel, ETH Zürich
Daniela Nicklas, Stuttgart University
Dirk Timmermann, Rostock University
Lin Uhrmacher, Rostock University


13:30 — 15:00Presentations I
Ochsenfahrt, Ulf et al.: EPIC-Application and Experiences in Self-Organizing Smart Ensembles (full presentation)
Propp, Stefan; Buchholz, Gregor: A User Control Mechanism for Smart Appliance Ensembles (full presentation)
Klein, Michael; Schmidt, Andreas; Lauer, Rolf: Ontology-Centred Design of an Ambient Middleware (full presentation)
15:30 — 16:30Presentations II
Beckstein, Clemens et al.: A Reactive Architecture for Ambient E-Learning (full presentation)
Burghardt, Christoph; Giersich, Martin: Utilizing partial order planner for synthesizing probabilistic models (short presentation)
Reiße, Christiane; Heider, Thomas: A survey of Ambient Intelligence Projects (short presentation)
16:30 — 18:00Workshop Discussion

Important Dates

July 16th, 2007:Submission Deadline
August 6th, 2007:Notification of Acceptance
August 20th, 2007:Final versions due
Sep 10th, 2007:Workshop