Cheap digital cameras are readily available. They can be mounted on robots and used to build maps of the surrounding environment. However, these cameras suffer from several drawbacks such as a narrow field of view, low resolution and limited range due to perspective. These limitations can cause traditional approaches to Simultaneous Localization and Mapping to fail due to insufficient information content in the visual sensor data. This paper discusses these issues and presents a solution for indoor environments.
Trevor Taylor, Wageeh W. Boles, Shlomo Geva