A Real-time Door Detection System for Domestic Robotic Navigation

Abstract

Human beings use doors to access rooms and corridors, to know where they are, to know where they have to go, etc. Similarly, it would be quite useful for robots to be able to detect doors in order to accomplish more complex and flexible navigation tasks. Such a goal is even more desirable when domestic environments are taken into account. Moreover, if the human-robot interaction is considered, the use of this semantic information can be broadly used. In this paper we present a solid and complete door detection system which fuses data from an end-user camera and a laser rangefinder. By using both Haar-like features and the Integral Image, the computation time is significantly reduced when compared to other methods found in the literature. Extensive tests in real-world environments have been performed in order to prove the efficiency, robustness and real-time ability of our system.