No abstract

Abstract-The increasing number of interactive products in our environment leads to an accession of interaction between products and users. Nowadays, this interaction is described with different concepts than the interaction between the products. This makes it difficult to mutually replace users and products to create smart environments that are able to automatically perform tasks for the user, if suitable products are available. To advance the design of interactive smart environments and to bridge this gap, we introduce a concept for describing product-initiated interaction with users.

No abstract

Nowadays, the interaction between a product and the user is described using different methods than for product to product communication. This makes it difficult to replace users and products mutually to create really dynamical environments, capable of reducing the amount of interactions, if possible. To advance the design of interactive smart environments, we introduce a concept for describing productinitiated interaction with users and demonstrate, how they can be applied in practice. This allows to combine both, automated procedures and interaction with the user, to a new concept called interactionflows.

Abstract-Intelligent environments aim at supporting the user in executing her everyday tasks, e.g. by guiding her through a maintenance or cooking procedure. This requires a machine processable representation of the tasks for which workflows have proven an efficient means. The increasing number of available sensors in intelligent environments can facilitate the execution of workflows. The sensors can help to recognize when a user has finished a step in the workflow and thus to automatically proceed to the next step. This can heavily reduce the amount of required user interaction. However, manually specifying the conditions for triggering the next step in a workflow is very cumbersome and almost impossible for environments which are not known at design time. In this paper, we present a novel approach for learning and adapting these conditions from observation. We show that the learned conditions can even outperform the quality as conditions manually specified by workflow experts. Thus, the presented approach is very well suited for automatically adapting workflows in intelligent environments and can in that way increase the efficiency of the workflow execution.