AI camera brings real-time vision functions into the control loop
30 June 2025 
Copyright B&R
ABB’s machine automation division, B&R Industrial Automation, has announced a next-generation smart camera that brings advanced AI into the machine control loop. The camera, with built-in edge AI capabilities, enables real-time vision processing, dynamic model switching, and hybrid AI-rule-based inspection – all without halting production or requiring external hardware.
Austrian-headquartered B&R says the camera will offer machine-builders and manufacturers a compact, plug-and-play way of delivering fast, high-quality throughput, and cutting waste – even under highly variable production conditions.
“What sets our new smart camera apart is the ability to embed it in the control loop which enables tuning machines on the fly,” explains B&R’s chief technology officer, Florian Schneeberger. “It’s no longer just about seeing defects; it’s about reacting to them instantly.”
Another attraction is the camera’s flexibility. “Customers can switch models and combine AI functions in custom sequences, without interrupting production,” Schneeberger says. This will allow manufacturers to adapt rapidly to new product variants or types of defect, and to integrate AI incrementally alongside existing rules-based systems.
Unlike conventional systems that rely on external PCs or PLCs, the camera processes its image data locally, allowing real-time AI inference at the edge, powered by a processor that is up to 15 times more efficient than rival processors, according to the edge AI specialist, Hailo.
The camera supports a suite of AI-based vision functions from MVTec, including OCR (optical character recognition), anomaly detection, object detection, classification and semantic segmentation. These can be combined with traditional rules-based algorithms, allowing users to balance the flexibility of AI with the speed and precision of conventional vision.
This hybrid approach is said to be ideal for complex inspection tasks such as identifying types of product, detecting subtle defects, and verifying printed codes – all in a single pass and with a single device.
The smart camera integrates with B&R’s Automation Studio engineering environment. A low-code interface allows automation engineers to configure AI vision tasks, simulate results, and manage models without needing specialised expertise. Existing smart camera installations can be upgraded with a simple drop-in replacement.
“Automation engineers can deploy AI vision in hours, not days,” says Schneeberger. “And because the system is fully synchronised with machine control, it boosts overall equipment effectiveness without adding complexity.”
The camera offers a choice of three sensors from 1.3-5 megapixels, as well as seven different integrated lenses (in addition to c-mount variants). It is integrated into a machine network via a single cable with a M12 hybrid connector, which also supplies its 24V DC power. A second hybrid connection enables daisy-chain cabling with lighting.
B&R claims that the camera’s factory-calibrated optics and lighting will improve imaging repeatability by a factor of 10 or more, ensuring high-quality input for deep learning models. This will result in more accurate detection, fewer false positives, and better long-term performance.
The camera is designed to handle unpredictable variations in materials, lighting and positioning. Its ability to perform multiple inspection tasks in one pass simplifies hardware requirements and shortens cycle times, while delivering measurable gains in productivity and quality.
B&R: LinkedIn
