Knowledge-Based Reasoning on Semantic Maps

Robotic systems should have a deep and specific knowledge about the environment they live in to properly interact with people and effectively perform the requested tasks. To this end, a suitable representation of the environment is needed, including both metric spatial information and semantic representations of locations and objects of interest. In this paper a representation of spatial and environmental knowledge, as well as a method for reasoning on it are presented. More specifically, the representation method is designed to properly integrate metric information about the environment and semantic information provided by the user, allowing for an effective knowledge-based reasoning. The result is a qualitative high-level representation of the environment that embodies all the knowledge required by a robot to actually reason on it and execute complex tasks.