Freitag, März 31, 2023

Machine Vision

The Machine Vision Portal

3D Machine Vision Makes Robotics More Efficient

  Presently in the field of robotics, only a low percentage of challenges is solved by a machine vision solution. In these cases, engineers develop the software themselves, thus, they barely revert to standard software. However, MVTec’s HALCON now offers the relevant possibilities for 3D machine vision in robotics like 3D measurement and 3D positioning. Therefore, HALCON not only provides fast and reliable operators for typical 3D methods, but also eases fast teaching and calibration, which are essential features for cost-effective robotics. From the robotics point of view, this is a clear push in efficiency. For 3D machine vision under HALCON, there are different possibilities. If the setup is based on two cameras, two variable stereo techniques can be disposed. The first one is important for precise mold measuring. It determines the depth of each single object point based on both transformed images via disparity (distance of corresponding pixels). However, often it is sufficient to only recognize single points, angles, or edges as selected features. In addition to increased speed, it shows the advantage that this second technique can deliver sub-pixel or rather sub-millimeter accurate position data to the robot.

Image of a soldering on a circuit board. Effected by different illumination (photometric stereo), HALCON is able to transform the object into a 3D reconstruction. 

3D methods also exist for one-camera -setups. After 3D camera calibration of the setup, HALCON reconstructs the object’s 3D position in world coordinates (method of monocular reconstruction) if the coordinates of the selected attributes of an object are known. A similar method (3D position of circles) makes use of circles in objects like drill holes, e.g., appearing in the camera image as ellipses. By the analysis of these ellipses, the 3D position of the objects can be reconstructed. Furthermore, a reconstruction is possible via depth by focus: With the height adjustment of the camera, the 3D-conditions of the object can be determined. Photometric stereo acquires multiple images with illumination from different orientations. The depth information is reconstructed by making use of the reflectance features of the object (s. fig.).

Beyond these special 3D methods for robotics, HALCON offers all classical features for machine vision such as measuring, matching, OCR, data- and bar code reading, and blob analysis. For in-situ quality control by robotics, also these classical machine vision methods will be employed more and more, e.g., for identification of components during bin picking in automotive engineering or for weld seam control. In the future, MVTec will also develop more and more 3D methods for robotics.

HALCON runs on PC as well as on special platforms (HALCON Embedded) used in robotics.

Hannovermesse (Hanover, 2006, April 24-28):     * MVTec: Hall 17, D60
* “The Unerring Dartboard” under HALCON, Hall 17, D30 (VDMA)

AUTO­MATICA (Munich, 2006, May 16-19):
* MVTec: Hall B1, 208
* “The Unerring Dartboard” under HALCON, Hall B1, 330 (iwb)