Automatic acquisition of spatial layout based on co-occurrence relationships between objects

Abstract

The layout of objects in space strongly affects people and robots. In a previous study, the authors defined layout as the structure of co-occurrence relationships among objects in space. They proposed a method to apply it to spatial mapping technology. In this paper, we propose a method for automatically obtaining layout information in space by automatically estimating the presence or absence of co-occurrence relationships in objects based on sensor data collected by a robot. Estimating co-occurrence relationships in objects based on geometric features resulted in the correct estimation of layout information with an accuracy of 72.1%.