“Angels and Devils” express a human’s dilemma. In this paper, we produced this expression with multiple robots and investigated how humans change their decision and behavior in a moral dilemma environment. We divided the participants into Neutral-Neutral and Angel-Devil robot groups and investigated the willpower of the two groups by the alphabet writing task. We also examined the impression of the robots by Godspeed questionnaires and the results of the questionnaire.

Emotional expressions are essential in augmenting a robot’s expression. Many robots with limited facial expression freedom, such as the Nao robot, can effectively express emotions using body movement. Previous studies have used a single robot and evaluated its expression. Multiple robots expressing emotion in their interactions may have a greater impact than robots expressing emotion only when interacting with humans. The relationships between these robots should allow them to develop more diverse modes of expression. However, it is unclear how people perceive relationships by observing robots’ emotional expressions. In this study, we applied every combination of four characteristic body emotion expressions (Sadness, Fear, Pride, and Happiness) based on Russell’s circumplex model to robots. Furthermore, we investigated how the relationships were evaluated in an online video-based experiment. The results show that the relationships between the two robots are influenced by each robot’s body emotional movement and can be interpreted using the valence-arousal model.

“Angel and Devil” expression has been used to cause a dilemma in various animations and comics. In this video demo, we implemented this expression with multiple robots. We supposed a scene in daily life, a student carrying a backpack and two robots riding on both shoulders to encourage the student to make decisions. This research concept can be the novelty use case of personal AI robots. In future works, the decision-making methodologies suggested by behavioral economics, such as nudging and boosting, can adopt to human-robot interaction by using multiple robots.

A variety of chatbots were developed to elicit people’s emotional responses and manage the emotions of people. These chatbots often use text-based emotional sentences. However, there are few studies that chatbots use extremely simplified communications to induce some emotional responses. We investigated the effect of mood contagion and tried to induce positive or negative words from people on our web-based ”word chain game” system as the most simplified conversational style. We also defined the system’s emotion as making the system use words that have polarity, which means each word’s strength of positive or negative emotion. The result shows that users of the system using positive words were slightly influenced to use more positive words than the neutral or negative system.

As voice recognition technology develops, we talk to smart speakers and other voice assistants more and more frequently in everyday life. Many such speech dialog systems are designed to use honorifics. However, humans do not always use honorifics in everyday conversation. Thus, we consider that voice assistants need to use other forms of expression in addition to so-called polite language. In this study, we conducted a comparative experiment to investigate how the speaking styles of smart speakers affect human listeners. In this experiment, a participant talked to a smart speaker to make travel plans. We analyzed the results of questionnaires. We found that a smart speaker's friendly speaking style influenced the participants' impressions and thus shaped the development of good relationships between humans and agents. Our findings indicate that people regard agents as friends if they speak in a friendly manner.

We modeled a robot's approaching behavior for giving admonishment. We started by analyzing human behaviors. We conducted a data collection in which a guard approached others in two ways: 1) for admonishment, and 2) for a friendly purpose. We analyzed the difference between the admonishing approach and the friendly approach. The approaching trajectories in the two approaching types are similar; nevertheless, there are two subtle differences. First, the admonishing approach is slightly faster (1.3 m/sec) than the friendly approach (1.1 m/sec). Second, at the end of the approach, there is a `shortcut' in the trajectory. We implemented this model of the admonishing approach into a robot. Finally, we conducted a field experiment to verify the effectiveness of the model. A robot is used to admonish people who were using a smartphone while walking. The result shows that significantly more people yield to admonishment from a robot using the proposed method than from a robot using the friendly approach method.