3D Map Building for Mobile Robots Using a 3D Laser Range Finder

Abstract

3D map building for mobile robots under cluttered indoor environmentsremains a challenge. The problem lies in two aspects: map consistencyand computational complexity. A novel method to finish such a taskis proposed. The system employs a special designed 3D laser rangefinder, which is built on the basis of a 2D laser scanner, as theenvironment-perceiving sensor. The registration of 3D range imagesis realized by localization on a so-called ceiling map, which isthe set of intersecting lines between the ceiling and the walls inthe room. By matching the calculated local ceiling map with the globalone, accurate sampling positions without suffering from accumulativeerrors can be obtained. A consistent 3D map can then be built onthe basis of it. The experimental results demonstrate our success.