Optimal fitting of objects based on range images - a planning approach assisted by local and global measurement matching

This paper presents a planning approach for optimal fitting of objects based on range image data of a laser scanner (triangulationbased). The approach includes two steps: local and global fitting. Local is based on matching each measurement to a nominal feature resulting in geometrical transformation parameters for each one of them. While several measurements are taken on a rigid body global aspects of the object must be taken into account such as: parallelism, or alignment of features. Therefore a global matching is introduce dregarding all the measurements simultaneously. Planning is presented as an optimization problem which results in the best fitting parameters. The approach is demonstrated by fitting a 2D object, and through planning an experiment in which robotic assembly of a car door is assisted by range image data.