3D Imaging for a Safe Belt - Sorting plastic parts
Lutz GmbH is testing components for belt restraints with a 3D image processing system. STEMMER IMAGING supplied the company with the required imaging components and provided support in implementing their first 3D system.
At first glance, the plain white plastic part being checked and sorted in the inspection machines of Lutz GmbH in the German Swabian town of Alfdorf does not reveal its functionality. And yet it has an important and potentially life-saving function as part of a belt restraining system in motor vehicles. "The surface areas of these plastic elements must be 100% levelled in order to ensure proper functionality", explains Managing Director Günther Lutz with regard to an important requirement for the inspection. "Furthermore, it must be absolutely certain that the parts do not have cracks or burrs", Michael Müller adds, who was the responsible technical expert for the installation of the testing equipment.
It was difficult to check the plastic parts manually and at a throughput speed of 40 parts per minute we quickly realised that we needed a powerful machine vision system. The framework conditions for developing the inspection machines were rather challenging as planarity testing is one of the more demanding applications in machine vision and the white plastic caused a lot of light scattering making the image analysis more difficult.
"We had already looked at a few 2D imaging approaches as well as compact and intelligent 3D systems, however, we were not able to find something that was up to the task", Mr. Müller remembers. "We only found a solution when we dared to consider triangulation-based 3D machine vision for inspecting the parts. Neither me nor any of my colleagues in this company ever had any exposure to this technology, however, the supportive STEMMER IMAGING experts enabled us to develop a suitable system within a short timeframe."
Success with 3D Imaging
The system Lutz implemented for inspecting the plastic parts consists of the following machine vision components: A linear laser from the manufacturer Z-Laser illuminates the components that are moved through the inspection machine on a conveyor belt. A 3D camera from Automation Technology is installed over the belt at an angle to the laser line and captures the laser profiles according to the triangulation principle. Common Vision Blox, the image processing library from STEMMER IMAGING, is part of the image acquisition and provides the driver for capturing the 3D images.
The 3D images captured this way are subjected to metric calibration und are finally converted to 3D point clouds. Subsequently, 3D Express executes a level fitting as well as a back projection of the 3D image to a 2D gray scale in the fitted level. This 2D image still contains the test object coordinates from metric calibration and can be evaluated with conventional 2D tools in the next step. A gray scale level corresponds to a height of only a few µm.
Evaluation of the images converted this way is done with the Sherlock image processing software from Teledyne DALSA, which Mr. Müller is particularly impressed with: "If a company such as ours is implementing mechanical engineering projects, the combination of 3D Express and Sherlock is the perfect platform. The matched tools offer great flexibility and permit conversion of different data structures as well as simple calibration." Mr. Müller created the testing sequence for inspection of the safety belt elements himself using Sherlock and has been excited about this technology ever since: "I have a keen interest in 3D machine vision and I would like to use it in other machines in the future wherever it makes sense."
In the described system, the parts are transported through the machine on the conveyor belt before being ejected at the end of the conveyor path. Depending on the evaluation by the image processing system, they are ejected into containers for good and bad parts, for which a maximum number can be defined in the system. As soon as the maximum number of pieces for a container is reached, the machine stops and indicates to the machine operator that a container must be replaced.
"We have only the best experience with this installation meaning that all parts that were evaluated as good parts were indeed good", Managing Director Lutz describes the outcome of the development. With an estimated 6 million parts inspected so far, he has all the reason to be proud of the performance of the installation.
Fast Implementation of the Installation
Mr. Lutz and Mr. Müller were delighted by the collaboration with STEMMER IMAGING. "I was particularly happy with the fast implementation of the image processing part of the installation. After our first contact with STEMMER IMAGING we identified the most suitable components for our application with a feasibility study in the Puchheim laboratory and thus found the final combination", is how Mr. Müller describes the development phase. He also said that the support relating to 3D machine vision was extremely helpful as this technology was entirely new to him. "Only three months passed between ordering the selected imaging components and the commencement of series operation – that exceeded my expectations", Mr. Müller said. An additional reason for the fast implementation was the use of 3D Express and Sherlock, which is a matched combination of two software solutions with graphical user interfaces. This enables simple configuration of 3D image capturing and the metric calibration using the image analysis software, which does not require any programming skills. The result is simple and fast access to solutions that would not be possible without 3D imaging.
The department manager used the extra time to optimise the installation for testing other objects. "Being a sorting contractor, Lutz GmbH faces the special challenge of having to adjust the machines to new or previously used test objects within a short timeframe", Mr. Müller explains. An important goal is reproducibility: "The results must still be correct if the machine was converted to and used for another testing task in between."
Mr. Müller already has ideas for further development of this and other installations: "We are planning an extension of the system and, for example, to install more ejection options for different good parts. This would help us to implement fine sorting of similar products." Mr. Müller also aims to increase the speed of the installation to 60 parts per minute and to implement a faster adaption to other test objects with various optimization measures.
During the collaboration with STEMMER IMAGING, Mr. Müller was also introduced to a new system that, if technically sensible, he would like to use in future installations: "I had a closer look at Trevista during a training session in Puchheim and the approach convinced me. I also see a lot of potential for our installations in this."