RI: Small: Collaborative Research: Preference Elicitation and Device Scheduling for Smart Homes

点击翻译按钮获取中文摘要

A home automation system (HAS) is an automated system that controls a home's smart devices with the objective of improved comfort, improved energy efficiency, and reduced operational costs. The home automation problem involves determining user preferences and constraints, scheduling smart devices to satisfy user constraints and minimize energy costs, and proposing to users schedules for these devices and responding to users' changes. As interactions between users and their HASes are likely to be limited, it is not practical for the system to elicit all preferences and constraints prior to scheduling user and home devices. The overall goal in this project is to design an HAS that can find acceptable solutions for users within a bounded number of interactions between the user and system. Towards that end, this project investigates techniques that elicit a small number of preferences and constraints that are key in finding good solutions for the user. If users are unhappy with the proposed solution, they can provide additional preferences, which will guide the search for a new solution. Through investigations of the home automation problem, this project will make the necessary foundational contributions to the field of cyber-physical systems, especially in the algorithmic techniques that take into account interactions with human users. Findings from this project will improve the design of future systems and guide the development of commercial HAS, which has the potential to impact future smart and connected communities.In order to find acceptable solutions within a bounded number of user interactions, this project will (1) model the bother cost of interacting with human users; (2) improve preference elicitation techniques by incorporating bother costs in their optimization; (3) use a portfolio of matrix completion algorithms and leverage their diversity to approximate unelicited preferences and their corresponding degrees of uncertainty; (4) design robust and scalable algorithms that find schedules based on uncertain preferences; and (5) propose visual and verbal interfaces that take into account bother costs while presenting and explaining those schedules to users.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.