Quality and process management are of great importance for companies in the automotive industry. Immense demands are placed on original equipment manufacturers and suppliers to ensure high quality and guarantee continuous improvements, waste reduction and error minimization. Particular attention is also paid to the interior of a car, as it makes a major contribution to the 'feel-good atmosphere' for the driver. To ensure perfect seating comfort in the vehicle, vehicle manufacturers place correspondingly high demands on the foams to be processed: they should be adaptable and flexible, resistant and durable, but also have a good feel and look. However, high quality standards must always go hand in hand with cost-effectiveness.

With this in mind, Adient, an American manufacturer of car headrests and armrests, recently commissioned an automated trimming line for these products at its Slovenian plant in Slovenj Gradec. The aim of this system was to increase the number of cycles and accuracy in the production of small foamed parts, reduce waste and thus increase the effectiveness of the manufacturing process.

Two robot cells

The solution for the precise trimming of armrests and headrests was developed and integrated together with the French company Gips Vision. This supplier of 3D systems for foam production, based in Marseille, designed two robot cells for this purpose, which literally put the finishing touches to the product after it has been manufactured on the production line.

In the first cell, the 3D data of the foam products is first captured using an 'Ensenso N20' 3D camera from IDS. Once the test object has been identified, a robot guides the camera around the product, creating a 3D point cloud of the foamed headrests or armrests. This creates an exact 360° image of the foam product, which the processing unit compares with a predefined 3D model. The exact position of the trimming curve on the actual foam surface is determined point by point.

The actual reworking begins in the second cell. The robot integrated there trims the product using the data determined in cell 1 and cuts the foam with a grinding tool. During this trimming process, any burrs or foam defects that the driver could feel on the contact surface are removed.

The system works to an accuracy of 2/10 mm and can be used around the clock. It can finish up to 120 foam parts per hour. "For our customer Adient, this is the first step towards automating the trimming of their foam parts," explains Thomas Derrien, Innovation Project Leader at Gips Vision. "The robots with integrated image processing make daily work much easier, as they avoid repetitive, tedious manual tasks. At the same time, they ensure a noticeable improvement in quality."

Triangulation creates exact 360° images of the headrests or armrests being viewed.

© IDS

Gips Vision opted for the 'Ensenso N20' camera from IDS. The main reasons for integrating this model were its high speed of up to 10 Hz and the volume covered with a camera resolution of 1.3 megapixels. With these characteristics, it was possible to achieve the accuracy of 2/10 mm required by the customer. In addition, the robust, compact aluminum housing of the 3D camera with screwable GPIO connectors for trigger and flash is particularly suitable for the harsh environment of a production hall.

All 'Ensenso' 3D cameras work according to the 'Projected Texture Stereo Vision' method, which is modeled on human vision. Each model uses two CMOS sensors and a projector that projects auxiliary structures onto the object to be captured. In this case, these structures are displayed on the foam parts in order to increase the accuracy of the surface image.

The two CMOS sensors of the 3D camera used in cell 1 view the respective foam part from different positions. Although the image content of both camera images appears identical, they show differences in the position of the headrest or armrest being viewed. Since the distance and viewing angle of the cameras and the focal length of the lenses are known, the associated software can convert these deviations into known lengths by triangulation. This makes it possible to determine the 3D coordinates of each individual object point and merge them into a 3D point cloud of the foam part to be processed.

The quality of the stereo vision process used is directly dependent on the lighting conditions and the surface properties (textures) of the objects. As foam has an easily recognizable texture compared to smooth or reflective surfaces, sufficient accuracy can be achieved by capturing just one image pair. The camera model used therefore enables fast image acquisition and image evaluation of the 3D point cloud. The camera series therefore impressed both the system integrator and the user.

From quality assurance to pick & place

The demands placed on manufacturers and suppliers in terms of quality and price are not only increasing in the automotive industry. Changes towards digital processes are also being driven forward in other sectors. The French company Gips Vision therefore not only develops market-oriented 3D applications for foam production, but also applies its expertise to similar applications. 'Ensenso 3D' cameras are proving their worth in a wide range of application scenarios, from quality assurance to pick & place.

Whether for the precise capture of special details on moving or stationary objects, even on shiny, dark and reflective surfaces, whether for use in multi-camera systems or for adaptive robot control: 3D vision and robot vision applications make day-to-day work in robotics and automated series production easier.

Author:
Silke von Gemmingen works in Corporate Communications at IDS Imaging Development Systems in Obersulm.