Modeling Buildings for Mobile Robots

In the near future mobile robots accomplishing such tasks as delivery of goods, assistance for handicapped people and surveillance will become increasingly available. The aim of this paper is to show the disparity that exists at present in modeling buildings for the purposes of two domains: Architecture, Engineering & Construction (AEC) and Mobile Robotics (MR). Digital product and process models developed in the AEC-domain give the user a comprehensive insight into the components and functions of the building but they neglect the needs of mobile robot operating autonomously inside the building. On the other hand, researchers in the MR-domain seem to be unaware of the results achieved in the AEC area. During the last decades the main effort in MR was devoted to the autonomous navigation. This resulted in efficient and robust procedures of acquiring floor maps from sensor readings. Further progress requires the robot to understand semantics of the surrounding world. Based on our previous research in the graph-oriented support of the AEC-design, we propose a digital model of the building that will enable the robot to recognize the functionality of particular spaces, to transform CAD-based floor plans into graphs reflecting topology and function of the building, to plan the execution of tasks posed by the human operator and to navigate inside the building.