Outline & Problem
Requirements for quality assurance in industrial environments force automated systems to increase in flexibility. Tasks still performed by people today like non-scalable manual inspection work should be decreased. This way a company can increase its scalability while handling thousands or millions of product items. Currently used automated quality inspection systems usually support only a few different item types, which is a huge draw back for agile market adoption of a production line.
Agile Quality Inspection
Today, many, many item inspection tasks are still performed manually by people. Even if several thousand parts need to be checked each day. Present inspection systems lack the flexibility to handle hundreds of different item types, which is why humans are still responsible to verify visual quality of parts (scratches, geometrical size, etc.). Our flexible QA solution relies on classical computer vision and deep learning perception to handle hundreds of different item types and to check them for defects and geometrical inconsistencies. You can use it for your in-line production quality assurance or commissioning processes, where you need to make sure, that the client receives exactly the ordered parts.
Approach & Methods
Facing the above need, GESTALT Robotics provides an advanced vision AI system, which is capable of inspecting hundreds of different object types. Combining classical computer vision and optimized deep learning algorithms on the software side, we enable high accuracy quality inspection and commissioning for your production flow. The solution is easily scalable, since the remote computing platform (edge or cloud) is capable to control several camera-based stations at once.
The processing enables real-time analysis being performed on embedded hardware or on powerful, remote instances saving hardware expenses. Apply the system on hundreds of stations and integrate the real-time inspection into your data warehouse or IoT infrastructure.
A common case in a production commissioning scenario is to verify, that all product items which are being shipped to the client are:
- free of defects
- assigned to the correct order
- checked to be put in the package out for delivery
Today, it is a very common case that a company does not produce a single type of product. For example, a bottle manufacturer offers bottles in many shapes and sizes. Thus, an automated inspection system handling all offered types of bottles is necessary.
Furthermore, maintenance is also required to be agile in terms of reconfiguration and extensions. If the manufacturer releases new types of products, using our AI solution, these can be easily integrated without help by a non-expert. The system only requires a few samples of a new product item to learn its shape, size and visual appearance.
Our quality inspection platform is deployed as a configured and calibrated sensor solution, optimized for the analysis of your product parts handling hundreds of different object types. Thanks to our data-efficient AI the system can learn new object types on-the-fly eliminating the reconfiguration overhead.
Drop us a message firstname.lastname@example.org or give us a call at +49 30 616 515 60. If you are dealing with manual, boring, repetitive tasks in your production flow, we are here to help you out.