Complete and detailed 3D scanning of cultural heritage artifacts is a still time-consuming process that requires skilled operators. Automating the digitization process is necessary to deal with the growing amount of artifacts available. It poses a challenging task because of the uniqueness and variety in size, shape and texture of these artifacts. Scanning devices have usually a limited focus or measurement volume and thus require precise positioning. We propose a robotic system for automated photogrammetric 3D reconstruction. It consists of a lightweight robotic arm with a mounted camera and a turntable for the artifact. In a virtual 3D environment, all relevant parts of the system are modeled and monitored. Here, camera views in position and orientation can be planned with respect to the depth of field of the camera, the size of the object and preferred coverage density. Given a desired view, solving inverse kinematics allows for collision-free and stable optimization of joint configurations and turntable rotation. We adopt the closed-loop inverse kinematics (CLIK) algorithm to solve the inverse kinematics on the basis of a particular definition of the orientation error. The design and parameters of the solver are described involving the option to shift the weighting between different parts of the objective function, such as precision or mechanical stability. We then use these kinematic solutions to perform the actual scanning of real objects. We conduct several tests with different kinds of objects showing reliable and sufficient results in positioning and safety. We present a visual comparison involving the real robotic system with its virtual environment demonstrating how view poses for different-sized objects are successfully planned, achieved and used for 3D reconstruction.
- Automation, Inverse kinematics, View planning, 3D Scanning
ASJC Scopus subject areas
- !!Chemistry (miscellaneous)