From labels to semantics: an integrated system for conceptual spatial representations of indoor environments for mobile robots

Abstract

We present an integrated approach for creating conceptual representations of human-made environments using mobile robots. The concepts represent spatial and functional properties of typical indoor environments. Our model is composed of layers which represent maps at different levels of abstraction. The complete system was integrated in a service robot which is endowed with laser and vision sensors for place and object recognition. It also incorporates a linguistic framework which actively supports the map acquisition process and is used for situated dialogue. In the experiments we show how the robot acquires the conceptual information and how it is used for situational and functional awareness.