go back

Detecting Availability to Facilitate Social Communication

Nils Einecke and Thomas Weisswange, "Detecting Availability to Facilitate Social Communication", ICRA Workshop: Exploring the Roles of Robots for Embodied Mediation, 2022.

Abstract

Modern communication has shifted strongly towards the digital domain. Technologies like chat applications and messengers provide means to connect with anyone at anytime. However, being always reachable can create stress and provides potential for social pressure due to expected timely responses. In contrast, the classical communication types via telephone is based on dedicated time slots for an ongoing communication, but suffers from the need to schedule or find a mutual opportunity to initiate this connection. Many Messenger applications allow the setting of a status to provide information about inappropriate times for initiating a communication, but constant updating of these states is often not feasible. We propose a framework for automatic detection of communication-availability and for signaling this to dedicated interaction partners using a video-mediated communication embodiment. In contrast to typical automatic status setting, our approach does not require a person to use or be close to its device, as it is not relying on device usage but rather on recognizing everyday activities and situations. Our algorithmic framework analyses situations using mainly image and audio features that are perceived through the communication device and indicates opportunities to start a communication to a possible partner. Within this framework, we compare a number of different machine learning techniques for classifying availability in a simple, real-world test dataset. The best approach is able to predict availability with 83% accuracy and presence of communication partners with an accuracy of 95%.



Download Bibtex file Download PDF

Search