Cognex Corporation has launched the In-Sight® 6900 Vision Controller, a new system aimed at expanding the use of artificial intelligence in manufacturing inspection and automation.
In a media release, the company said the product is designed to give manufacturers greater flexibility in configuring machine vision systems while improving processing performance at the edge for industrial environments.
According to Cognex, the modular system allows manufacturers to select and combine cameras, lenses and lighting components based on specific inspection requirements, rather than relying on fixed-configuration systems.
The company said this approach is intended to reduce system compromises and improve adaptability across different manufacturing applications.
“As manufacturers accelerate their adoption of AI, they want solutions that are both powerful and easy to deploy,” said Matt Moschner, President and CEO of Cognex. “The In-Sight Vision Controller delivers exactly that, combining NVIDIA’s edge AI processing with Cognex’s modular hardware and proven vision tools,” he said.
Cognex said the system is powered by NVIDIA Jetson technology, enabling high-capacity AI processing at the edge so manufacturers can run advanced inspection workloads without external PCs or complex distributed computing setups.
The company said this is intended to simplify deployment while supporting higher-performance industrial applications.
The In-Sight Vision Controller introduces updated AI tool modes designed to address complex manufacturing inspection challenges, including variability in part size, defect detection and inconsistent production conditions.
Cognex said the system includes few-sample classification models that require as few as 10 to 20 training images, as well as pixel-level segmentation tools intended to improve detection accuracy on complex surfaces.
Cognex said the platform also integrates with its OneVision software environment, which enables manufacturers to develop and manage inspection applications across multiple sites.
The company said this is intended to help standardise inspection processes and accelerate updates across production lines and facilities.
The company said the system delivers up to 157 TOPS of AI performance using NVIDIA Jetson hardware, allowing multiple high-resolution models to run in parallel.
It added that NVIDIA TensorRT integration supports real-time inference designed to keep pace with high-speed manufacturing lines operating at microsecond-level timing.
