Acting as if Being Aware of Visitors' Attention Strengthens a Robotic Salesperson's Social Presence

Robotic salespersons have been adopted by many shops. However, due to the weakness of their social presence, they are easily ignored by visitors. Thus, we focused on such robots' behaviors that can strengthen their social presence in a real-world shop. In this research, our goal is to develop a model that can obtain visitors' replies to the robot's utterances. For this reason, we conducted a laboratory experiment and two field experiments. In the laboratory experiment, we found that the robot's ability to express that it can understand the degree of visitors' attention to the robot is important for improving its social presence. Afterward, we developed an engagement estimation model (EEM) based on the findings obtained from the laboratory experiment. We defined visitors' engagement as the probability that the visitors will reply to the robot's utterances. The EEM estimates the engagement from the visitors' real-time nonverbal data. In a laboratory experiment, the situation of being in the laboratory can make the interaction between the participants and the robot unnatural. Therefore, we developed a model based on data of visitors' nonverbal cues in the field experiment and examined whether the model works effectively in the real-world environment. As a result, the automatic greeting mode based on the EEM facilitated the participant's reply to the robot. Thus, we considered that this is because the automatic greeting mode could make a decision of the robot's next behavior precisely. It can be considered that the visitors may feel as if the robot could understand their degree of attention to the robot when the robot greets the visitors based on the automatic greeting mode. Therefore, the automatic greeting mode based on the EEM worked effectively in a real-world environment and could strengthen its social presence.