GUESs: Generative modeling of Unknown Environments and Spatial Abstraction for Robots

Representing unknown and missing knowledge about the environment is fundamental to leverage robot behavior and improve its performance in completing a task. However, reconstructing spatial knowledge beyond the sensory horizon of the robot is an extremely challenging task. Existing approaches assume that the environment static and features repetitive patterns (e.g. rectangular rooms) or that it can be all generalized with pre-trained models. Our goal is to remove such assumptions and to introduce a novel methodology that allows the robot to represent unknown spatial knowledge in dynamic and unstructured environments. To this end, we exploit generative learning to (1) learn a distribution of spatial landmarks observed during the robot mission and to (2) generate missing information in real-time. The proposed approach aims at supporting planning and decision-making processes needed for robot behaviors. In this paper, we describe architecture modeling the proposed approach and a first validation on a mobile platform.