In a modern manufacturing plant, there are a wide range of machines continuously moving, making, and assembling products. The days where machines simply follow instructions are gone; today, they think, decide, and execute with detailed precision.
Vision inspection systems use high-resolution cameras, sensors, and sophisticated image processing software that allow for fast and detailed inspection of products as they move along the production line. The vision system captures images, processes them in real-time, and checks for any inconsistencies or defects. AI implementation plays a pivotal role as it enhances the system’s efficiency in image processing, real-time analysis, and data analytic capabilities to recognize production process trends. It can better evaluate object texture and shape variances against the product’s expected dimensions than other methods. Due to the advancing automation industry and implementation of AI, there is an increased need for higher-quality cameras, more powerful computing hardware, and advanced software to effectively utilize its capabilities.
Identifying defects is just half of the process. Modern factories are interconnected with wireless connectivity. Vision systems, which were historically wired systems, are integrating system-on-modules with wireless connectivity to establish a secure wireless connection with a cloud server. Once the vision system identifies the defective part, it packages the parts data, often with information like product ID, timestamps, and snapshots, in formats such as XML or JSON to be transmitted to the cloud.
In the cloud, the data doesn’t just sit there. It’s stored into its tracking database where analytic tools sift through the data, spotting trends and patterns. Whether recognizing a sudden surge in defects or detecting recurrent issues within a batch, it provides valuable real-time insights for factories to identify the cause of problems and put them to a stop. The integration of Wi-Fi technology, which supports WPA3 and enterprise security, prevents data from being intercepted or from intruders gaining unauthorized access. Keeping the network up to date with the latest firmware and defense measures, such as firewalls, establishes multiple layers of security against cyberattacks.
Let’s look at a breakdown of a vision inspection system:
During vision inspection, it can be a challenge to add powerful components as these systems are usually small in size requiring space-saving components. The cameras must determine if an item is defective in real-time by analyzing data during powerful processing systems. This is where SOMs can help. Designed to be compact, SOMs fit seamlessly into the hardware of vision systems and its controller’s display. Laird Connectivity’s Nitrogen 8M Plus SMARC utilizes a Cortex-M7 microcontroller that runs a real-time operating system (RTOS) to handle computation-intensive tasks associated with AI algorithms to provide feedback of defective parts in real-time. With dedicated AI and machine learning capabilities delivering up to 2.3 TOPS, the inspection system improves over time as it is exposed to more data achieving higher degree of accuracy to recognize even the slightest inconsistencies.
During inspection, the camera captures and processes series of images by enhancing, filtering, and sharpening to ensure that the essential features are clear. The image is then analyzed using various techniques like edge detection and pattern recognition. Once a defective part is identified, the system sends the results to a central server, historically through Ethernet, however, since the release of Wi-Fi 5 with higher data rates and features like MU-MIMO (multi-user, multiple input, multiple output), wireless systems are being integrated due to better reliability and reduced latency. The Nitrogen8M Plus SMARC uses the Sterling LWB5+ Wi-Fi 5 and Bluetooth 5.2 radio, allowing easy data transmission, remote monitoring of inspection results, and over-the-air updates to run the latest and most optimized algorithms.
The next step of the process is vision inspection. The Nitrogen8M Mini SOM has built-in multimedia features, including 2D/3D graphics support and advanced video handling capabilities, which work smoothly with today’s high-definition inspection cameras thanks to its 4-Lane MIPI-CSI connection. Additionally, with memory options that can go up to 4GB LPDDR4 (Low-Power Double Data Rate) RAM and storage that can expand to 128GB eMMC, the device ensures quick data processing and ample space for saving data. Laird Connectivity offers an off-the-self single-board-computer (SBC) of the Nitrogen8M Mini with the necessary peripherals for quick time-to-market.
Lastly, as the system captures, analyzes, and processes the data, the operators need to be notified of potential issues. Integrating Bluetooth modules into vision system’s hardware can help notify the necessary personnel on the floor. Utilizing a module like BL5340 series, featuring Bluetooth LE 5.2, enables communication to provide feedback to operators about inspection results. If a bad batch needs to be segregated from the finished products or the production machinery needs to be adjusted, these tasks can be delegated right away to nearby personnel.
In the digital transformation era, vision inspections systems utilize advanced SOMs and wireless communication to enhance automation, precision, and operational efficiency in the evolving manufacturing landscape.