Fitting CAD models to point clouds

  • Subject:Applied robotics of the automation industry
  • Type:Master Thesis
  • Supervisor:

    Witucki, Linus

  • Links:Tender
  • Implementation of an algorithm for fitting parameterizable CAD models to measured point clouds


The production of wiring harnesses for the automotive industry is largely carried out by hand using state-of-the-art technology. The individual cables are laid on a board and then fixed in harnesses. The cables are laid in forks (see Figure 2) for fastening on a cable forming board (see Figure 1). The correct position of these forks on the cable forming board is therefore decisive for the accuracy of fit of the finished cable harness. Errors in the position of forks therefore lead to deformed wire harnesses that cannot be installed in the target vehicle. For this reason, point clouds of the cable moulding board are recorded and digitized. For complete digitization, CAD models of the individual forks are to be merged iteratively to form an overall model of the wire harness.


The aim of the thesis is to create a parameterizable CAD model of a standard fork. This is then to be fitted to a measured point cloud. The parameters of the fork should represent production-relevant variables such as the distance between the tines and the cable feed-through point. The results of the fitting are the production-relevant parameters and the transformation matrix of the fork.

Helpful prior knowledge

The following prior knowledge is helpful for working on the thesis:

  • Optimization
  • image processing
  • ROS
Figure 1: Wiring harness
Figure 2: Fork