In order to reliably handle objects in dynamic everyday environments, robot systems must be able to recognize and localize objects. New 3D image processing technologies can be used to search specifically for feature points, which are combined and stored to form a model. This means that objects can be recognized even in changing light conditions and partial occlusion. The three-dimensional object recognition system can do even more: it can deduce the class or category of the object from the combination of geometric shapes. Thanks to the combination of object recognition and classification, the robot can then independently 'learn' specific objects or general object classes or be intuitively trained on them. 3D image processing also opens up new possibilities ...