Emotional Expression with Long-Term Interactions in a Linear Affect-Expression Space
Haeun Park
Research Keywords
Human-Robot Interaction;
Social robot;
Affective model;
Interaction design
Figure 1
Facial expression of PoomGo, DECS Lab, 2020
Research Statement
The expected growth rate of personal service robots that can directly assist people is high [Bartneck 2004, Gemma 2006].
Also, Social robots that are used as pets among robots with high growth potential. By comparison, however, social robots are not as popular as AI speakers or smartphones, although many companies have put them on the market so far. Reasons for the failure include high prices, non-categorized products, simple interactions, and lack of design [Hoffman 2019].
Studying robots that can have deep interaction with humans is my main research topic. Social robots that exist today quickly lose interest in people due to the absence of long-term interaction. People do not regard pets as toys, but as friends or family. I think long term interaction will make social robots like pets. It will be largely divided into three stages. Recognize a person's feelings in the first step. Recognized results determine the robot's emotions in an emotional model that enables long term interaction. The determined emotions will be output from the display or voice.
Figure 2
The proposed human robot social interaction framework architecture. [Lee 2019]
"Figure 2 shows the proposed framework architecture for the human robot social interaction. The architecture is composed of sensory part, recognition part, language part, memory part, communication part, control part, cloud applications.
The robot receives user’s face, voice, touch as input, and makes social interaction possible through the appropriate reaction output. Sensory part collects input data using image, sound, and touch sensor." [Lee 2019]
Among them, I am studying memory parts that are involved in emotional exchange between robots and users. This memory stores emotional episodes with people in the past, enabling long-term interaction.
3
"Moxie is designed to help all kids improve their social-emotional and communication skills through relatively short daily interactions." [Evan 2020].
Moxie is an educational social robot for children.
Moxie has long-term interaction features that will help education without boredom for growing children.
These Moxie-like results are the final result of the research I want. Although it is difficult to implement everything, we will implement an emotional model that is long-term interaction.
Figure 3
Moxie, Embodied [Evan 2020]