Autonomous Mobile Robot Navigation: Consideration of the Pedestrian's Dynamic Personal Space

Abstract

Recently, the demand for service robots has increased and these robots have already been implemented in our daily life. Many service robots move autonomously in an environment wherein humans and robots coexist using a function to arrive at a given destination. Some studies have proposed autonomous mobile robots that can effectively avoid a pedestrian in the coexisting human-robot environment. However, these studies did not consider the dynamic personal space of pedestrians and create discomfort or a feeling of psychological pressure in the pedestrian. Therefore, herein, we measured the size of a pedestrian's dynamic personal space while interacting with an autonomous mobile robot and modeled this space using the property that the space changes with relative velocity and relative acceleration. We created an artificial potential method considering this dynamic personal space and realized robot navigation that can direct a robot to a given destination while avoiding a pedestrian. In a simulation experiment, we confirmed that our navigation algorithm functioned accurately and that the robot did not invade the subject's dynamic personal space. The latter was also confirmed in a practical experiment with our prototype robot using measured distance and a questionnaire.