Master Thesis MSTR-2021-16

BibliographyIvanova, Desislava: Dynamic personalized home automation rules based on multi-user feedback.
University of Stuttgart, Faculty of Computer Science, Electrical Engineering, and Information Technology, Master Thesis No. 16 (2021).
73 pages, english.

With the increasing use of smart devices in homes, the need for Home Automation Systems (HASs) has grown. A feature that most systems provide is automation rules. These rules are often defined in a format that is not easily understandable for all users and which may require prior knowledge or experience with HASs. Additionally, the rules are static, meaning that they need to be manually redefined by the user. This can be problematic since a single condition cannot be used in all scenarios. This can be improved by having dynamic rules that can be changed using user feedback. The aim of this thesis is to examine how automation rules can be presented to the users of HASs in a human-readable form, and how automation rules can be dynamically updated after receiving explicit user feedback. Respectively, these two problems are also defined as the research questions of the thesis. The first step of the research regarding the first research question is to come up with an approach that can translate an automation rule into human-readable text. For this purpose, existing research is examined. Afterwards, a solution is proposed and its implementation is thoroughly discussed. The proposed translation works with rules defined with the help of Node-RED. A triggered rule is parsed using a grammar created specially for the rules. The parsed rule is added to an Abstract Syntax Tree (AST) that is then used to form a comprehensible text that describes the essence of the rule. For the purpose of answering the second research question regarding the dynamic actualization of automation rules, components that allow static rules to be dynamically changed using explicit user feedback are proposed. Additionally, a system is build with those components. The system uses existing components like Home Assistant (HA) and Node-RED and builds up on them, adding the components implementing the functionalities responsible for presenting the user with the rules, gathering his feedback, and updating the rules. Additionally, the translation approach proposed in the thesis is also included. The component responsible for the presentation of the rules and the collection of feedback is implemented as an Android application. The application presents the users with the text of a rule when it is triggered and gives them the ability to provide feedback on aspects of the rule. When the feedback is submitted it is used to update the corresponding rule in Node-RED. The proposed system is then evaluated using different use cases. The use cases have different levels of complexity, meaning that they include a varying number of sensors and actuators. The evaluation of the system based on the mentioned use cases shows that the system is capable of translating automation rules into human-readable text and is able to dynamically update these rules with the feedback of the users. The evaluation also examines the limitations of the presented solution.

Full text and
other links
Department(s)University of Stuttgart, Institute of Architecture of Application Systems
Superviser(s)Aiello, Prof. Marco; Setz, Brian
Entry dateJuly 27, 2021
   Publ. Institute   Publ. Computer Science