KEYENCE has launched the VS-G Series, a controller-type AI vision system built around a 32-core processor the company says runs up to 13 times faster than its VS-L/S/C Series platforms when running a 5-megapixel AI segmentation tool. The headline addition is what KEYENCE is calling a "world's first": dedicated image storage integrated directly into the controller, enabling continuous capture of every image a line produces, including images of parts that passed inspection.
Until now, storage constraints on high-speed production lines forced engineers to archive only defective-part images, discarding the visual record of accepted parts. The VS-G addresses this with a proprietary compression algorithm that cuts image file sizes by a factor of 100 while maintaining quality equivalent to the original. What's new here is that quality managers can back-test any configuration change against a stored archive of historically passing images, confirming a new detection parameter won't generate false positives before it goes live.
Setup relies on Auto Parameter Tuning and Auto Image Selector, two tools that reduce manual configuration. An operator selects a reference image, highlights the defect area on the interface, and the controller calculates and applies inspection parameters automatically. A Settings Verification Function then cross-checks configuration changes against the stored archive to confirm adjustments don't alter judgment on previously accepted parts.
A single VS-G controller supports up to eight simultaneous camera connections, drawing from a lineup of more than 30 area-scan camera models. AI Segmentation and AI Detection Fine Mode require the optional VS-AD1 module. Applications cited by KEYENCE span PCB appearance inspection, wafer SEMI character reading, sealant bead and gap measurement in automotive assembly, and cargo verification in wire mesh logistics trays.
The VS-G launch comes as inline AI inspection moves from pilot programs toward production requirements for audit-ready data trails. Growing customer and regulatory demands for complete part histories, particularly in automotive and semiconductor supply chains, have exposed the absence of OK-part image archives as a gap that conventional vision controllers leave open. KEYENCE's integrated storage closes that gap without adding external storage servers to an existing line.
