Unknown objects segmentation and material classification for separation

This work address the issues and difficulties related to the design of an autonomous robot capable of performing unknown objects segmentation and material classification task. One of the most challenging aspects of such a system is that the robot has to segment unknown objects from a complicated scene and in the proximity of other objects. In this paper, illuminations of different frequencies are projected from the robot, providing additional information about the scene compared to conventional segmentation techniques. By using multiple light sources and material's reflectivity we were able to identify true edges and separate segmented unknown objects. To classify and gather information about the unknown segmented objects we introduce a novel material classification technique using static electricity charge sensing. Our proposed approaches do not require prior models of target objects and assumes no previously collected background statistics.