//Logo Image
PDF Version



1. SEBASTIAN, Y-L. HSU. Talking to the home: IT infrastructure for a cloud-based robotic home smart-assistant. Gerontechnology 2018; 1l(Supp/):102s; https://doi.org/10.4017/gt.2018.17.s.099.00 Purpose Dialog with robots using natural language has becoming an  important  trend  of the development  of companion  robots.  A 2017 study by Google and Peerless Insights showed how users feel comfortable to include a smart assistant in their home1. In this study, participants also manifested that having a cloud-based smart assistant system integrated to their homes allowed them to ease their daily tasks. Even more insightful is the fact that users felt that they were talking to a "friend" or "another person" while interacting with the  smart assistant. The  2018 consumer electronics show (CES) showcased a long variety of robots2, many that offered a cloud-based smart assistant system  integrated  in them. There are 3 possible types of dialogs, information  based, task based and chatting. The purpose of this paper is to describe the design of an effective  IT  infrastructure  that  allows  a  robot  to interact with the older adults and receive control home environments. Furthermore, such IT structure should be easily installed in existing home environments. Method The core concept of the design is, the entire home as

a robotic platform. The integration of a robotic user interface with cloud-base smart assistants at home allows users to have a focal point of interaction when interacting with the entire home. Figure 1 presents the overall infrastructure of an loT home ecosystem. The loT home ecosystem consists of a team of devices that interact by listening and communicating with each other. Whilst the team of devices works together, the robotic user

interface is a gate of interaction to the users. This co anion robot embodies an loT ready system, which uses the Google APK as the main brain for interaction. The information based and chatting are solely based on current Google technology. Figure 2 shows how task based on dialog is achieved. Users first state the command recognized by the API-AI agent created for the robot. This agent has been set to interact with users through natural language. This specific behavior creates the "illusion of life" in which users can command

their tasks in their own speech pattern. The natural language programming is scripted to sound and feel like a natural conversation whilst it waits for the specific command of the user to be spoken. Such  command  is converted to a web-hook, which now can be sent to either  the Google  home control  center  or the server  in which the loT devices are connected to (in this case AWS). These loT devices via the robotic user interface are now able to "listen" to voice commands and perform the tasks users request through the robot. Results & Discussion In this research, the role of  a companion  robot  is linked to the  loT  ecosystem  that allows  current and future devices to be controlled via voice commands.  A  prototype  companion  robot  was built and  linked to an loT living lab to demonstrate the feasibility. Initial user experience experiment  was conducted.  To make the task based dialog even more natural, each set of commands can have hundreds or thousands of iterations. Google's APK for smart assistants also allows robots to do machine learning and pairs specific commands to specific accounts based on location, user preferences and more allowing each home to be unique



1.   Gottsegen G, Miller J. CES 2018 gives a preview of our future robot overlords [ Internet]. 2018, January 9

[cited 2018 January 21]. Available from: https://www.cnet.com/pictures/ces-2018-robots-pictures/1/

2.   G, P. 5 ways voice assistance is reshaping consumer behavior.  2017,  August [ cited  2018  January  20] AvaiIable from: https://www.thinkwithgoogle.com/data-collections/voice-assistance-emerging-technologies/


Keywords: loT, human-computer interaction, Robots at home

Address: Yuan Ze University, No. 135, Yuandong Road, Zhongli District, Taoyuan City, 320, Taiwan;

E: jkings16@yahoo.com




Figure 1. Overall IT infrastructure of an lo T Home


Figure 2. IT structure of a natural language processing system for and loT home Ecosystem through a Robot



2018                                                                                         102s                                                               Vol 17, supplement