Co-speech Gesture Generation for Embodied Agents and its Effects on User Evaluation - Coverbal Synchrony in Human-Machine Interaction - page 237

seconds: “Hello, my name is Billie/Vince. In a moment you will
have the possibility to getting to know me closer. But first you will
be provided with a questionnaire concerning your first impression
of myself”. Subsequently, participants were asked to state their
first impression regarding their perception of the embodied agent's
personality in a questionnaire (T1, took approximately five minutes).
In the second phase, the embodied agent described a building with six
sentences which took approximately 45 seconds. Each sentence was
followed by a pause of three seconds. Participants were instructed
to carefully watch the presentation given by the embodied agent
in order to be able to answer questions regarding content and
subjective evaluation of the presentation afterwards. Immediately
upon receiving the descriptions by the agent, participants filled out
a second questionnaire (T2) stating their perception of the embodied
agent's personality at this point of time.
It turned out that with regard to perceived warmth , the agents were
overall perceived as warmer than at measuring point T2. In addition,
there was a significant interaction effect of point of measurement
and agent appearance: ratings of warmth decreased for the robot-like
agent between the two points of measurement, while ratings remained
constant for the human-like agent. In other words, participants rated
the robot-like agent as being warmer than the human-like agent at
measuring point T1, whereas ratings of warmth did not differ between
robot-like and human-like agents at T2.
Regarding perceived competence of the embodied agents, there was
a significant effect for the interaction of the point of measurement and
agent behavior. While gesture use was found to result in an increase
of perceived competence of the agents between the two points of
measurement, ratings of agent competence slightly decreased when
the agents did not use any gestures.
What can we learn from these findings for the design of (interactions
with) virtual agents? First of all, the results clearly show that there is a
second chance to make first impressions. In particular, with regard to
competence, employing virtual agents with gestures helps to increase
participants' ratings—independent of the agents' appearance. As a
consequence, we can advise to endow virtual agents with gestural
behavior to improve their perceived competence.
6. Conclusions
This chapter discussed the possibilities, approaches and effects of
modeling co-speech gestures for embodied agents. The GNetIc model,
described in detail, combines data-driven and rule-based decision