Mobile robot range sensing through visual looming

Abstract

Describes and evaluates visual looming as a monocular range sensing method for mobile robots. The looming algorithm is based on the relationship between the displacement of a camera relative to an object, and the resulting change in the size of the object's image on the focal plane of the camera. We have carried out systematic experiments to evaluate the ranging accuracy of the looming algorithm using a Pioneer 1 mobile robot equipped with a color camera. We have also performed noise sensitivity for the looming algorithm, obtaining theoretical error bounds on the range estimates for given levels of odometric and visual noise, which were verified through experimental data. Our results suggest that looming can be used as a robust, inexpensive range sensor as a complement to sonar.