Publication

SiAM-dp : An open development platform for massively multimodal dialogue systems in cyber-physical environments

Robert Neßelrath

PhD-Thesis Universität des Saarlandes 12/2015.

Abstract

Cyber-physical environments enhance natural environments of daily life such as homes, factories, offices, and cars by connecting the cybernetic world of computers and communication with the real physical world. While under the keyword of Industrie 4.0, cyber-physical environments will take a relevant role in the next industrial revolution, and they will also appear in homes, offices, workshops, and numerous other areas. In this new world, classical interaction concepts where users exclusively interact with a single stationary device, PC or smartphone become less dominant and make room for new occurrences of interaction between humans and the environment itself. Furthermore, new technologies and a rising spectrum of applicable modalities broaden the possibilities for interaction designers to include more natural and intuitive non-verbal and verbal communication. The dynamic characteristic of a cyber-physical environment and the mobility of users confronts developers with the challenge of developing systems that are flexible concerning the connected and used devices and modalities. This implies new opportunities for cross-modal interaction that go beyond dual modalities interaction as is well known nowadays. This thesis addresses the support of application developers with a platform for the declarative and model based development of multimodal dialogue applications, with a focus on distributed input and output devices in cyber-physical environments. The main contributions can be divided into three parts: - Design of models and strategies for the specification of dialogue applications in a declarative development approach. This includes models for the definition of project resources, dialogue behaviour, speech recognition grammars, and graphical user interfaces and mapping rules, which convert the device specific representation of input and output description to a common representation language. - The implementation of a runtime platform that provides a flexible and extendable architecture for the easy integration of new devices and components. The platform realises concepts and strategies of multimodal human-computer interaction and is the basis for full-fledged multimodal dialogue applications for arbitrary device setups, domains, and scenarios. - A software development toolkit that is integrated in the Eclipse rich client platform and provides wizards and editors for creating and editing new multimodal dialogue applications.

Projekte

Weitere Links

thesis_-_final.pdf (pdf, 21 MB)

German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz