Authors
Saad Elbeleidy
Venue
Colorado School of Mines PhD Dissertations
Publication Year
2024
Research on robots that assist through social interaction, termed Socially Assistive Robots, suggests a
promising future for their adoption. Past research often focuses on the robots’ design and functionality or
the assistance provided by the robots and its efficacy, often forgetting the people behind these robots, the
care wizards, who are key to the success of these robots. This document presents a collection of work that
centers the experience and needs of the people critical to the success of Socially Assistive Robots; not the
technologists who build these robots, but the domain experts without whom these robots could not succeed.
The degree to which a person is uninvolved in the control of a robot is termed the Level of Robot
Autonomy. We begin by reviewing past research on Socially Assistive Robots, examining the motivation
for that work and selection of robot autonomy, and comparing it to guidelines on selecting robots’ level of
autonomy. We find a plurality of research in pursuit of autonomous Socially Assistive Robots without
explicit motivation for that work. However, guidelines from research on selecting robot autonomy suggest
that more teleoperation is appropriate.
This work includes several studies aimed at understanding how care wizards operate Socially Assistive
Robots in practice. We analyze archival data of children’s after-school programs where robots were used
and find themes in robot verbalizations and consistent patterns in how care wizards use robots. We
interview experienced care wizards about their time delivering therapy with robots, uncovering that
therapy is a dual-cycle process, and identifying several broad and specific needs relating to care wizards’
robot use. We run group usability tests with domain specialist novice care wizards and discover their
barriers to adopting robots in their practices; primarily technology complexities and lack of domain-specific
content to be used. We re-examine our interviews with experienced care wizards and find a high likelihood
of necessary but often overlooked work, or invisible labor, required to author content for care wizarding.
Our need-finding efforts conclude with an extensive list of detailed recommendations for designing tools for
care wizards, many of which highlight the importance of authoring.
To design tools for care wizards, we provide a systems-level design and examples of specific capabilities
that may support care wizards. We recognize that work on End-User Development, or tools designed for
general users to author complex behaviors, can be leveraged in designing authoring tools. We apply service
blueprinting, a service design method that makes work visible, to teleoperated Socially Assistive Robots
and generalize them as tools of End-User Development for Teleoperated Robots. We conclude with
examples of interfaces that may support care wizards in note-taking, reflecting, and authoring. This work
is merely the start of designing for care wizards and we encourage future work in this space.
