Rikei has launched sales of POLYGON VISIO, a semi-packaged edge-AI system for visual inspection on manufacturing lines. The system combines a camera, lighting, a compact edge computer and a stainless-steel enclosure, with shipments scheduled to begin on October 1. The stated entry price is ¥5 million, varying with configuration.
POLYGON VISIO is intended to reduce the engineering work that often precedes an automated inspection project. Instead of requiring a customer to assemble vision hardware and computing from separate components, the product arrives as a configurable inspection unit. Rikei planned the product, while partner M&M Solutions handles development, manufacturing and support. The companies are targeting factories that want to automate part of a manual inspection process without rebuilding an entire production line.
What is included
The product is described as a semi-package because its core hardware is assembled but can still be designed and customised for different objects and industrial requirements. Rikei says the edge computer permits high-speed inspection without a required connection to the company network or a cloud service. It can therefore operate as a standalone inspection device, which may simplify deployment where production data must stay on site.
| Item | Disclosed detail |
|---|---|
| Product type | Semi-packaged edge-AI visual-inspection system |
| Included elements | Camera, lighting, edge computer and stainless-steel enclosure |
| Starting price | ¥5 million; final price varies by configuration |
| Shipment date | October 1, 2026 |
| Support | Three months of post-installation support included |
Why local inference changes the rollout
Keeping inference at the inspection station can reduce dependence on network latency and cloud availability. It may also narrow the path through which production images leave the line. That does not automatically make a deployment secure or compliant: buyers still need to review device access, software updates, model files, logging and any remote-support connection. The announcement confirms that a network connection is not required for operation, not that every deployment is isolated by default.
The included three-month support period is meant to help customers learn the AI and expand it to multiple target objects. That addresses a common gap in inspection projects: a model that works during setup can lose usefulness when lighting, materials, packaging or defect patterns change. Rikei has not published accuracy, throughput, false-reject rates or supported object limits in the launch release. Those figures will depend on the configured line and should be established in acceptance testing.
A semi-packaged system also changes responsibility. The supplier can standardise the camera, enclosure and compute platform, while the factory still defines what counts as a defect and how a result affects the line. That division should be explicit in the commissioning plan. Without representative samples and a controlled change process, even consistent hardware can produce inconsistent operational decisions when products or tolerances change.
The launch reflects a broader move to place AI inside operational workflows. Hitachi’s Nokia Drone Networks offering similarly packages hardware, connectivity and software for field operations. In a different production context, Brainstorm Suite 7’s AI production tools integrate automation into an established operator environment rather than creating a separate task queue.
What factories should test
A buyer should begin with representative good and defective samples across shifts, lighting conditions and production speeds. Acceptance criteria should separate missed defects from false alarms and record how performance changes as new object types are trained. Teams should also confirm who owns the images and model data, how updates are validated, what happens when confidence is low, and whether the line can continue safely if the inspection unit stops.
Rikei plans to display POLYGON VISIO at the Automatic Identification Systems Expo in Tokyo from September 9 to 11. It also says a simplified POLYGON VISIO NEO is planned, but no launch date or price for that model was provided. The current news is the sales launch of the main system and its October shipment schedule.
FAQs
Does POLYGON VISIO require cloud access?
No. Rikei says the edge computer allows it to operate without connecting to a company network or cloud platform.
How much does it cost?
The announced starting price is ¥5 million, with the final cost changing according to the selected configuration.
Is the system shipping now?
Sales have been announced, and shipments are scheduled to start on October 1, 2026.
Sources: Rikei’s release, Weekly BCN+ and Dempa Times.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
