Subjective Impression Perceived from Agent that Can Recognize Discrepancy in Visual Perception of Situation

In this study, we focus on the ability of human symbiotic intelligent systems to recognize discrepancies in perception of the situation, especially visual perception, with humans. Even when they share the same space, the system’s perception of the situation and that of the humans sometimes do not match. To truly understand and predict the other’s behavior in such a situation, the system must recognize that the other is unaware of a particular fact. Behaviors based on such ability are expected to make a good impression on users and contribute to establishing a desirable relationship with them. However, this expectation is still a hypothesis and has not been verified. In this study, we carry out a subjective evaluation experiment to verify the hypothesis. For the experiment, we created a personified agent that interacts with the user and can recognize discrepancies in visual perception of the situation using virtual reality technology. We also designed an interaction scenario that includes a situation where a discrepancy in perception occurs. Using this agent and scenario, we carry out the experiment to investigate how the agent’s behavior in the situation affects the user’s subjective impression. Experimental results imply that an agent’s ability to recognize discrepancies in the perception of the situation gives the users a positive impression.