This paper presents the novel paradigm of a global localization methodmotivated by human visual systems (HVSs). HVSs actively use the informationof the object recognition results for self-position localization and forviewing direction. The proposed localization paradigm consisted of three parts: panoramic image acquisition, multiple object recognition, and grid-basedlocalization. Multiple object recognition information from panoramicimages is utilized in the localization part. High-level object informationwas useful not only for global localization, but also for robot-object interactions. The metric global localization (position, viewing direction) wasconducted based on the bearing information of recognized objects from justone panoramic image. The feasibility of the novel localization paradigmwas validated experimentally.
展开▼
