Incremental version space merging approach to 3D object model acquisition for robot vision

Abstract

A concept learning algorithm is developed, which uses the visual information generated by a virtual receptor in a robotic system (e.g. symbolic image segments) to create learning examples. Its goal is to detect similarities in the training data and to create an appropriate object model. The version-space, intended to describe the possible concept hypotheses, is generated by a novel IVSM-ID algorithm, the incremental version space merging with imperfect data, that deals with partly imperfect and noisy training data—a common problem in computer vision systems. The generated model takes the form of a graph of constraints with fuzzy predicates. The approach is verified by learning concepts of elementary surface and solid primitives on base of segmented RGB-D images, taken for various light conditions and for different exposure times.