Prof. Dr. Susanne Biundo-Stephan
The technology transfer project T3: "Do it yourself, but not alone: Companion-Technology for Home Improvement" was approved by the German Research Foundation at the end of September 2016. In the project research findings of the SFB/TRR 62 "Companion-Technology for Cognitive Technical Systems" are put into practice and transferred into industry. In collaboration with the global research and advance engineering of the Robert Bosch GmbH a prototypical Companion-system for home improvement support will be developed.
For the realization of the assistant the Institute of Artifical Intelligence and the Institute of Communications Engineering offer their expertise as well as technologies developed within the SFB/TRR 62. The work will be conducted in three sub projects. As for the Insitute of Artifical Intelligence this involves the projects Automated Planning as well as Knowledge Representation and Reasoning under the direction of Prof. Dr. Biundo-Stephan and Prof. Dr. Birte Glimm, respectively. Both projects originally emerge from the project A1 of the SFB/TRR 62. In addition the Institute of Communications Engineering under the direction of Prof. Dr. Dr.-Ing. Minker contributes research results from the project B1, Assistive and Adaptive Dialogue Management.
The assistance system supports its users with a detailed step-by-step guidance in the course of executing a home improvement project. The guidance is computed at runtime and explains in-depth what the user has to do. The corresponding instructions are illustrated on several abstraction layers in between the user can switch whenever needed. This allows to take account of different types of users with distinct expertise knowledge, since it is not necessary to provide experienced do-it-yourselfers with the same level of abstraction for the step-by-step instructions as for beginners. In addition the assistant also instructs how to handle the applied tools.
All this happens depending on the present state and the available attachements. For example, if a device is used several times with different attachements, the assistant will instruct the user to remove the battery first for security reasons before changing the attachement. Further the system knows which attachements are needed for what do-it-yourself project, because for instance drilling in wood requires other attachements than drilling in stone. All the knowledge necessary for the assistant exists in a formal specification and hence can also be explained if needed. That means if a user is unsatisfied with only instructions on what he or she should do (e.g. with which attachement tools) but also wants to know why, this can be explained in natural language at runtime. In addition to this basic functionality, research is conducted on how the assistant can deal with execution errors as well as the possiblity to enable users to deviate from the original project plan. This will not only individualize the presentation of the project guidance, but also the whole do-it-yourself project itself and its realization respectively.
Furthermore do-it-yourselfers will get the opportunity to specify own do-it-yourself projects, in whose realization the system also can assist. The workflow will then be provided on a community platform for further usage even without the assistant.