Triangulation & Hough Transform Based Fusion of Sonar Data for Mobile Robotics

Abstract

Polaroid sonar sensors are still one of the most widely used sensors in mobile robotics. In this paper a novel sensor fusion scheme called THF (triangulation Hough fusion), which combines the advantages of TBF (triangulation based fusion) and CIHT (cluster inhibiting Hough transform), is presented. The idea of THF lies in that a modified Hough transform is employed on the points extracted by TBF, not on the raw sonar data directly. THF algorithm applied on clustered and noisy sonar data could reliably detect both point and line segment features of the environment. We demonstrate the effectivity of THF algorithm by extracting geometric features in a simulated indoor environment, in which a mobile robot travels with pedestrians moving around casually. Experiments demonstrate that THF is an effective and efficient scheme for detecting both point features and line segment features, especially in the structured environment.