Group Norm Awareness of Mobile Robots in Virtual Space Multiagent Simulation

Abstract

The emergence of social robots has created new opportunities for coexistence with humans, provided they can imitate human-like social behavior. Numerous studies have explored the navigation of mobile robots in the context of socializing robot behavior within human society. In a space where humans and robots coexist, the robot’s positioning, that is, the way it occupies space, should be natural and human-like. Personal space theory, which is a model for maintaining distance between people, categorizes interpersonal distance. The distance maintained can be dynamic, which depends on the situation and context regardless of whether a person is comfortable or uncomfortable. In a society where humans and robots coexist, it is required to adapt to the behavior of humans who maintain such a context-dependent distance. Little consideration has been given to the point at which the robot continues positioning itself as a group member, which is the focus of this study. We propose a navigation method that introduces a concept of “near neighbors” for this purpose and assess the path generated by this method through a multiagent-based scenario in a virtual space. The simulation results demonstrate that our proposed navigation method adapts the implicit norms of distance and the experimental scenarios. In our future study, we intend to evaluate our proposed method, including a subjective evaluation by humans.