Object recognition for robot work piece identification by the use of a laser scanner combined with Fourier transform

Determination of the three-dimensional shape of objects for their recognition is an important problem. One important tool for object identification is vision. As a complement to vision laser scanner technique can be used. Differently shaped objects scatter the laser light differently. The scattered light is detected and converted to an analog voltage signal. To characterize the signals and to be able to classify the objects out of the signal we use Fourier transformation. The Fourier transformation gives a few parameters that can be used for classification. If we want to use the whole spectra for classification the statistical principal component analysis gives a good result. This method is a rather simple and inexpensive way to complement conventional object recognition systems, e.g. in robotics applications