SICK Introduces Inspector83x with AI-Powered Quality Control
The SICK Inspector83x 2D vision sensor brings out-of-the-box AI machine vision inspections to common inline inspection tasks in high-speed production.
June 5, 2024
At a Glance
- SICK’s Inspector83x can be used by non-specialists to configure high-precision AI inspections.
- Pre-installed with the SICK Nova foundation software, the Inspector83x is designed to offer simplicity out-of-the-box.
- Advanced users can expand custom developments with Lua programming and HALCON.
SICK has introduced a vision inspection platform that draws on artificial intelligence. The system is designed for configuration without the need for programming. The no-stress teach-in capability of the SICK Inspector83x can be used by non-specialists to configure high-precision AI inspections at full production speeds. By showing the camera just a handful of examples, initial inspection results can be up and running in a matter of minutes. Any product design or batch changes can be quickly added in the same way by an operator from the production line.
Simplicity and Power in High-Speed Production
With up to 5MP resolution and built-in illumination, the SICK Inspector83x is an all-in-one performer. With its quad-core CPU and high-speed data transfer over industrial networks, the Inspector83x is designed to process AI inspections directly on the device, and at faster speeds than its predecessors. Its highly accurate and fast response, with no need for external machine control, is expected to be particularly useful in FMCG production applications. Typically, up to 15 inspections per second can be accomplished reliably for machine vision tasks such as defect and anomaly detection or classification.
Image courtesy of SICK.
With the SICK Inspector83x, manufacturers can solve machine vision inspections with ease while depending on highly reliable results in applications such as consumer goods manufacture, food and beverage, automotive, and packaging applications.
The SICK Inspector83x offers complex inspections of medium to large scenes, including evaluating products with unpredictable features and verifying complex assemblies, as well as OCR/OCV reading and verification. Additional product roll-outs during 2024 include color imaging for inspection of vivid features suitable for color sorting, defect detection and quality assurance.
AI Machine Vision for Non-Specialists
Pre-installed with the SICK Nova foundation software, the Inspector83x offers simplicity out-of-the-box. Neither specialist machine vision expertise nor lengthy problem-solving is required to specify an inspection successfully, according to SICK.
Using a standard PC connected via the camera‘s USB-C port or network interface, users can follow the intuitive interface to present examples to the camera in actual production conditions, then train and execute the inspection. As little as five examples may be needed. Combining the AI function with conventional rule-based tools, e.g. to add a simple measurement value, allows inspections to be configured in the most pragmatic way.
The SICK Inspector 83x can eliminate conventional machine vision complexity whenever changes to product designs or packaging are needed. Rather than having to call on a machine vision specialist or external consultant to problem-solve and specify complex rule-based inspections, non-skilled operators can simply add a new product example and the camera will learn by itself.
For complex scenes with many more examples and large datasets, users have the option to access the computational power of the SICK dStudio cloud service to train their own neural network, which can then be exported as a small file to run on the Inspector83x.
With SICK Nova, advanced users can also expand custom developments further using Lua programming and HALCON. The SICK cloud-based dStudio service also offers the unique advantage of colleague collaboration and data management, still with user-simplicity at its core.
High-Speed Data Transfer with Versatile Industrial Networking
Once set up, the image inference is carried out directly on the SICK Inspector83x and results are output to the machine control as pass/fail results or sensor values. The Inspector83x is optimized for rapid data transfer in industrial networks with dual ports for EtherNet/IP™ or PROFINET integration.
A dedicated high-speed Gigabit Ethernet port provides the capacity for high-resolution image data transfer as well as data logging or TCP/IP integration. A built-in export function can output customized configurations for common PLC types at the push of a button.
The SICK Inspector83x has seven inputs and five outputs. The on-board delay and queue capability precisely calibrates the camera‘s image outputs based on either time or encoder inputs to activate connected machine controls e.g. to trigger a reject pusher.
Although the SICK Inspector83x operates well as a stand-alone device, a broad range of accessories is available where required for the installation. A dedicated port for illumination can connect to external light sources such as backlights or bar lights.
A near infrared variant will be available later in 2024. Using a standard C-mount threading, customers can fit a range of optics from SICK, as well as having the freedom to fit specialized third-party lenses for complex inspections.
Read more about:
Supplier NewsAbout the Author
You May Also Like