Powerful eyes for agricultural and food robots

Humans heavily rely on their sight for performing most of their tasks. Apart from classical RGB camera vision, which is a logical choice when aiming to replace human sight in robotics applications, there are several other optical measurement techniques which can provide very valuable information to the robotic system. Distance information can be obtained from stereovision, but time of flight measurements have some clear advantages which have made these the preferred choice for several tasks in agricultural machinery automation. In the complex, unstructured environments with variable environmental conditions which are typical for agriculture, additional spectral information obtained with multi- or hyperspectral imaging systems can give considerably better object classification than the limited information contained in the red, green and blue bands of an RGB camera. Apart from discrimination between different object classes, the diffuse reflectance spectra also contain valuable information on the quality of the individuale products in terms of their composition and microstructure. As the interaction of light with biological tissue is the result of a complex interplay between light scattering and light absorption, different measurements concepts which allow to separate the information on the light scattering and light absorption are also discussed.