The Use of Social Sensorimotor Contingencies in Humanoid Robots
381
Given the ubiquity of selfies worldwide, we further expect that the general population will have sufficient experience with the task to make sense of the sensorimotor information displayed by the robot while taking a selfie. The rise of
popularity of selfies is currently the subject of scholarly discussion and motivations for taking a selfie have been cited as multidimensional (as they are taken
alone or in group settings) and a fun, creative outlet for self-fashioning and
self-expression [12]. Taking a selfie with another person may seem like a form
of enjoyable, shared self-expression, and as such, we want to investigate if this
interaction can yield a similar result when performed with a synthetic agent. The
task of taking a group selfie requires the social skills of recognising people who
can be grouped, getting their attention, establishing a connection, conveying
intention, gaining permission, maintaining connection and physical coordination
by uniting participants for taking the photo. Here, we wish to ground these skills
in sensorimotor contingencies while exploring the dimension of personality in the
context of taking a selfie and how they affect the transparency and outcome of
the interaction of taking a group selfie.
2 Methods
The main objective of this work is to make use of sensorimotor contingencies
to allow for mutual understanding between humans and robots and present the
system developed. We further seek to address the gap in research in context and
personality of social sensorimotor contingencies (socSMCs) in HRI, by evaluating
which higher-level behaviour as expressed through sensorimotor contingencies is
interpretable and preferable to humans during the task of taking a “selfie”. The
personalities we evaluate are introvert and extrovert, as they are most commonly
used in HRI, providing an easier comparison with other studies and are relevant
to the task of taking a selfie. We make the assumption that a robot’s use of
socSMCs will allow for effective flow and coupling necessary to take group selfies
regardless of the personality. We, therefore, hypothesise that humans will understand distinct combinations of contextually relevant sensorimotor cues in gesture,
gaze and spatial relation leading to effective coupling. We further hypothesise
that humans will prefer an extroverted robot due to the entrainment nature of
the task and that extroverted behaviour will lead to a greater number of selfies
taken and longer interactions.
To do so, we developed two models of interactions (introversion and extroversion) and analysed from a human-focused perspective, emphasising fluency
and interpretability of the interaction. To focus on the effects of solely sensorimotor channels of communication, we removed verbal interaction. The robot
relied solely on the use of sensorimotor information during interactions.
2.1 Setup
Drawing from literature in both HHI and HRI, the extroverted robot’s behaviour
as exhibited through social sensorimotor contingencies makes use of closer proximate distances, exhibits larger gestures, and acts as the “leader” (active) of the
381
Given the ubiquity of selfies worldwide, we further expect that the general population will have sufficient experience with the task to make sense of the sensorimotor information displayed by the robot while taking a selfie. The rise of
popularity of selfies is currently the subject of scholarly discussion and motivations for taking a selfie have been cited as multidimensional (as they are taken
alone or in group settings) and a fun, creative outlet for self-fashioning and
self-expression [12]. Taking a selfie with another person may seem like a form
of enjoyable, shared self-expression, and as such, we want to investigate if this
interaction can yield a similar result when performed with a synthetic agent. The
task of taking a group selfie requires the social skills of recognising people who
can be grouped, getting their attention, establishing a connection, conveying
intention, gaining permission, maintaining connection and physical coordination
by uniting participants for taking the photo. Here, we wish to ground these skills
in sensorimotor contingencies while exploring the dimension of personality in the
context of taking a selfie and how they affect the transparency and outcome of
the interaction of taking a group selfie.
2 Methods
The main objective of this work is to make use of sensorimotor contingencies
to allow for mutual understanding between humans and robots and present the
system developed. We further seek to address the gap in research in context and
personality of social sensorimotor contingencies (socSMCs) in HRI, by evaluating
which higher-level behaviour as expressed through sensorimotor contingencies is
interpretable and preferable to humans during the task of taking a “selfie”. The
personalities we evaluate are introvert and extrovert, as they are most commonly
used in HRI, providing an easier comparison with other studies and are relevant
to the task of taking a selfie. We make the assumption that a robot’s use of
socSMCs will allow for effective flow and coupling necessary to take group selfies
regardless of the personality. We, therefore, hypothesise that humans will understand distinct combinations of contextually relevant sensorimotor cues in gesture,
gaze and spatial relation leading to effective coupling. We further hypothesise
that humans will prefer an extroverted robot due to the entrainment nature of
the task and that extroverted behaviour will lead to a greater number of selfies
taken and longer interactions.
To do so, we developed two models of interactions (introversion and extroversion) and analysed from a human-focused perspective, emphasising fluency
and interpretability of the interaction. To focus on the effects of solely sensorimotor channels of communication, we removed verbal interaction. The robot
relied solely on the use of sensorimotor information during interactions.
2.1 Setup
Drawing from literature in both HHI and HRI, the extroverted robot’s behaviour
as exhibited through social sensorimotor contingencies makes use of closer proximate distances, exhibits larger gestures, and acts as the “leader” (active) of the