The software system MORAL (Munich Object Recognition and Localization) is used for the robust recognition of à priori known objects in monoculuar video images. Besides the object identification task, the rough spatial object pose relative to the camera position is computed. This pose can be used as an initial pose hypothesis for the subsequent pose estimation task. Using this information, an autonomous, mobile robot is able to manipulate objects (e.g. grasping) and to navigate through indoor environments.