12 May The eye of artificial intelligence – Computer Vision
Computer vision is a sub-area of artificial intelligence and deals with the fact that computers can extract information from data. The computer is able to see, identify and analyse visual data and to provide an appropriate result. This data includes videos, images, scans and multidimensional data. The goal is for the computer to adapt human abilities and perform them with equal efficiency. An example would be the ability to reproduce human vision so that images can be captured and understood electronically.
Tasks of Computer Vision
One of the most important tasks is object recognition. This means that the computer can identify, locate and classify objects in images.
If the computer is able to divide objects into discrete categories like “boat”, “house” or “dog”, this is called classification. Computer vision can also be used to identify specific people. This is called face recognition. The computer is able to identify and classify several persons within a picture. Object localization is another useful skill. The position of an object in the image is recognized. This helps to estimate the distance of an object to the camera.
For example, when driving a car, a computer can extract important data from an image and provide it to the driver. The computer is able to detect objects next to and on the road, such as pedestrians, traffic lights, signs and others. The next step is for it to distinguish between the individual data it captures and act accordingly.
The application area of Computer Vision is wide. It ranges from industrial image processing systems, i.e. the recognition of objects on an assembly line, to the field of artificial intelligence.
Computer vision can be used to make work steps easier and clearer. Combining augmented reality with computer vision, it is possible to perform work steps on an assembly station. This station records the steps of the work process and indicates errors. In real time you get feedback if a step was executed correctly or if it needs improvement. In addition, the parts to be machined are highlighted in colour so that the user always knows which section is to be machined. VISCOPIC has built such a station for Porsche to make it easier for them to assemble parts.
Assembly process validation in the Porsche Pilot Center
Among many other technologies, VISCOPIC uses computer vision for process innovation, digitization and optimization of industrial processes of companies.
Together with Porsche, VISCOPIC has developed a prototype that helps technicians to assemble rotor-lamination packages in the right way. For electromagnetic reasons there are different ways in which they have to be assembled. Normally Porsche uses the Poka-Yoke principle, but it is not possible to implement it in this special component design. Therefore, the technicians are faced with the challenge of assembling the many parts in the correct order.
The developed prototype is a digital poka yoke using 2D computer vision and augmented reality to support the employees. A prototype assembly workstation with automatic component recognition was developed for this case. Here a camera detects the geometric differences and extracts geometric features from the CAD data. With the VISCOPIC Computer Vision Setup, sheet metal packages can be scanned individually and compared with the CAD originals. The information is forwarded to the AR software, which instructs the operator and thus counteracts incorrect installations. If an incorrect part is used, it is detected by the camera and informs the operator. This approach can be applied to many other component types. Based on the generic solution, this setup can be used to detect a variety of components based on CAD components.
Your company can also enjoy the benefits of computer vision. Contact us at firstname.lastname@example.org and we will guide you with your use case into digitization.