No Innovation Without Scientific Curiosity

Our research and development team constantly develops our hardware and software products further. But you can only be innovative if you always broaden your horizon and are open for new impulses. No innovation without scientific curiosity.
We at SAC are aware of this fact and therefore actively support soon-to-be researchers and engineers by offering them the opportunity to write their final thesis at our company.

SAC supports bachelor and master students of computer sciences, electrical engineering, optical technology and machine vision, production engineering, or similar subjects in working on an exciting, current topic related to machine vision. “We offer continuous support from topic identification to writing up the findings, high practical relevance, the use of state-of-the-art technology in an industrial environment for research or engineering, as well as a friendly work atmosphere – everything you need for a successful final thesis.” says Alexander Piaseczki, deputy head of the development department.

The latest example of a successful thesis completed at SAC comes from a student of the Karlsruhe Institute of Technology who dealt with the use of artificial intelligence in machine vision. The task was to check how useful neural networks can be in determining the position and rotation of an industrial component.
The task was performed with a convolutional neural network. An artificial neural network is a network of artificial neurons that is modeled after natural neural networks, i.e. the network of nerve cells in the brain and spinal cord.
The aim was to get the neural network to determine the position and rotation of an industrial component from 2D gray-value images in such a way that the obtained data could subsequently be transmitted to a robot that could grab the component correctly and supply it to the next production step (a “pick-and-place” task).

To achieve this goal, the neural network had to be trained under supervision. By “backpropagating” errors, the network learned step by step. With each new sequence, the result approximated the nominal value.

By using different approaches and continuously adjusting parameters, the neural network could eventually be trained so well that the target of reaching an accuracy of ±0.5 mm in position determination and of ±5° in rotation determination could be reached. This showed the potential of neural networks for the use in machine vision.

“We gladly take up this input from research and we will surely delve deeper into the topic of artificial intelligence in machine vision. We're also looking forward to further input from smart minds. We're always ready to assist interested students with our expertise, our technical know-how, and an open ear. Of course, we profit at the same time from the fresh ideas and new perspectives – it's a win-win situation for all involved.” according to Piaseczki.

Are you interested in writing your final thesis at SAC?
Write to us at