Autonome 3D-Objektmodellierung durch aktive Szenenexploration mit einem UR5 Greifarm


The 3D modeling process is often time-consuming. The object, that needs to be modeled, must be examined from all sides in order to create a 3D copy of this object. This bachelor thesis deals with a process of 3D modeling of small objects. The scaning process is autonomous. The object is examined with a depth camera that is attached to a UR5 gripper arm. During the examination, the pointclouds from the object surface will be recorded with the camera and saved. This thesis deals first with the solution of the registration problem and then with the conversion of registered pointclouds to a solid 3D object. This bachelor thesis shows different methods of solving these problems. The pointcloud registration is done with different variants of the ICP algorithm. The 3D objects from the pointclouds were made using the Poisson Surface Reconstruction and the Alpha Shape algorithm. The presented methods are explained in detail together with their mathematical basics.