Bin picking, robot guidance, distinguishing dents in components; all are examples of where 3D vision excels, as was demonstrated at Stemmer Imaging's 3D vision technology day that took place on 29 February. And a lot of the applications benefitting from adding a third dimension are found in the automotive sector, a detail that chimed well with the setting for the event which was held at Mercedes-Benz World in Weybridge, Surrey, UK.
Measuring the contour of a car door, for instance, for robotic sealing of welds relies on a 3D profile in order to thread the sealant nozzle in the gap between car body panels. This robotic sealing system was developed by VMT, a turn-key solutions provider based in Mannheim, Germany. The sealant nozzle is 2mm wide and had to fit in a 3mm gap between door panels, requiring a very exact positioning method. VMT used a laser triangulation sensor to monitor the gap width and calculate the door contour in order to thread the nozzle in between the panels.
VMT has also developed systems for best-fit positioning of car body panels, a solution that's currently in use in Daimler factories. Triangulation sensors measure the gaps and flush fit, the data from which allows a robot to assemble the panels correctly.
According to a market study from the Fraunhofer Allianz-Vision, 84.5 per cent of the German automotive supply industry employs 3D measurement techniques to some extent. The traditional approach in manufacturing cars is to use a contact profilometer probe, which, as Mark Williamson, director of corporate market development at Stemmer Imaging, noted, offers opportunities for non-contact optical techniques as a replacement for contact methods.
In his presentation, Dr Frank Grünewald, a project manager at VMT, also showed how time-of-flight sensors are used to find the edges of pallets for robotic depalletising. Similar sensors were also used for bin-picking tasks, such as the removal of break discs from storage containers. The discs are lying in random orientations in the container and the sensors identify their placement and position in order for the robot to choose the most accessible from the pile.
So much of 3D imaging revolves around uses and properties of light, be that a laser line in triangulation, detecting deformation in a light pattern to acquire a 3D shape, or measuring interference or the time it takes for light to return from an object to give depth information. Dr Christoph Wagner from OBE presented its Trevista lighting solution, which illuminates the object from four different angles to create a 3D image from 2D shade images, while Dr Khaled Aljasem from Z-Laser identified laser stability as an important consideration when implementing a laser triangulation system.
A lot of these 3D techniques however did depend on the material under inspection as to how effective they are; highly reflective or transparent surfaces are difficult to inspect with most 3D imaging techniques. The 3D surface profile of glass, for instance, would be difficult to generate. And as Williamson of Stemmer Imaging pointed out, ‘3D vision doesn’t necessarily solve everything’. However, for certain applications like robot bin picking, adding a z-axis is critical, and understanding the capabilities of different 3D vision techniques and when they should be applied will all help with the technology’s uptake.
Other companies presenting at the event included Aqsense, Automation Technology, LMI, GFMesstechnick, and GOM UK.
