Architecture considerations for interoperable multi-modal assistant systems

Creating multimodal assistant systems supporting the intuitive interaction with technical infrastructures of the everyday life is one important goal of current HCI research. Building such systems is a substantial challenge - not only with respect to the individual concepts and algorithms that are required at the various levels of multimodal interaction processing, but also with respect to the overall system architecture. Especially, when we try to address systems that can be extended dynamically and that are not built by a single vendor. We will look at some the challenges of creating architectures for such systems and we will outline the solution approach our research group is currently developing. The work we present is part of the EMBASSI-project, a joint project with 19 partners from industry and academia that aims at establishing an interoperable system infrastructure for multimodal and multimedia assistance systems.