Edge AI for Industrial Vision Applications

Bring intelligence closer to the point of action with Visteon's D6Sigma Edge AI IoT Box portfolio. Designed for industrial vision deployments of any scale, these purpose-built platforms enable manufacturers to capture, process, and manage visual data across production environments using a single, scalable edge infrastructure.

Edge AI Box Portfolio

Built to support a range of deployment sizes and operational needs, Visteon's portfolio provides a scalable foundation for industrial vision applications. From focused installations to large, multi-site environments, each solution is optimized for different camera densities and coverage requirements while leveraging a common platform architecture for simplified deployment and management.

 

Edge-AI Pro

4-8 Cameras

A compact edge AI platform for smaller deployments across a limited number of production lines, workstations, or operational areas.

Edge-AI Pro Max

8-16 Cameras

A scalable solution designed for larger facilities, factories, warehouses, and complex manufacturing environments requiring centralized management across multiple areas. 

Edge-AI NVR

16-32 Cameras

A high-capacity platform that combines video recording and AI processing for large-scale industrial environments, ideal for campus-scale operations and enterprise-wide industrial monitoring. 

Datasheet Get a quote

Why Manufacturers Are Moving AI to the Edge

Factory lines generate constant visual data — and nearly all of it goes unanalyzed. Defects slip through. Micro-stoppages go unrecorded. Safety violations happen undetected. The intelligence to act on these events exists. But until now, it hasn't been practical to deploy at the edge, where manufacturing actually happens.

Vision AI works in the lab. Getting it onto the production floor — reliably, at low latency, without cloud dependency — is a different engineering problem. Three issues have kept it out of reach.

Problem #1

Invisible Downtime

Micro-stoppages — halts lasting seconds to minutes — are individually too brief for any manual system to record. Collectively, they are one of the largest drivers of lost output.

Problem #2

Cloud Dependency

Sending camera streams to a cloud AI service introduces latency, privacy risk, and a single point of failure — unacceptable in a continuous manufacturing environment.

Problem #3

Integration Friction

Factory systems speak different protocols. Connecting a new AI layer to existing infrastructure typically requires months of custom engineering.

Closing the gap between AI potential and production reality

Existing vision systems weren't built for this. They handle single-function detection or require cloud connectivity that manufacturing environments can't tolerate. Visteon’s D6Sigma is different.

  • Edge-first — all inference on-device

  • Proven in Visteon's own manufacturing facilities

  • Open integration layer — REST, OPC-UA, MQTT

  • 60+ capabilities on a single platform

D6Sigma is Visteon's purpose-built answer to what factory-floor AI actually needs to be. Built on the CognitoAI™-IoT platform and Qualcomm Dragonwing™ AI-enabled processors, it runs real-time vision AI locally, at the point of capture — and connects directly to the systems manufacturers already operate.

Three layers that make it work together

To make factory-floor AI practical, three things need to work in concert: the right processor, the right software platform, and the right analytics layer. Here is how they map in D6Sigma.

  • Compute

    The AI compute foundation for the factory floor

    The D6Sigma Edge Device is the physical intelligence at the point of capture. Powered by Qualcomm Dragonwing™ AI-enabled processors, it processes up to 16 simultaneous camera streams locally.

    Two configurations are available: a multi-zone unit for factories and warehouses supporting 8–16 cameras, and a smaller configuration for 4–8 cameras suited to individual production cells or smaller facilities.

    Fleet-scale device management, A/B OTA updates, and ONVIF camera discovery are handled through Qualcomm Insight, keeping every deployed unit current without manual intervention.

  • Intelligence

    The software platform that reasons and acts

    CognitoAI™-IoT is the AI software layer that runs on the edge device, handling model execution, event classification, rule-based alerting, integration routing, and fleet management in a single on-box stack.

    Discovery Mode allows manufacturers to capture new video, label events with AI assistance, retrain models, and deploy them remotely — without data science expertise on the plant floor.

  • Analytics

    The operational intelligence that closes the loop

    Detection is only the first step. The Six Sigma Analytics Layer converts event streams into process metrics — making the invisible visible, and the measurable improvable.

    It provides 6S Workspaces with multiple SPC charts, SPC violation alerts, and KPI target setting. Event data is linked through Integration Studio to Andon, MES, and maintenance management systems using the protocols those systems already speak.

    Because all analytics run on-device, sensitive production data never leaves the facility.

D6Sigma handles two broad classes of factory intelligence

Every use case runs on the same platform — and manufacturers can extend it with their own models and event types.

  • Worker safety and PPE compliance monitoring
  • Changeover (SMED) and assembly-step verification
  • AGV and AMR traffic and material-flow monitoring
  • People counting and gesture detection
  • People counting and gesture detectionMachine status and predictive maintenance signals
  • Privacy face blur in regulated or sensitive environment

  • Quality inspection and defect / rework detection at line speed
  • Capture new video streams from any production area
  • Line monitoring, Andon integration, and micro-stoppage classification
  • Label events with AI assistance — no data scientist required on-site
  • Create or retrain models directly within CognitoAI™-IoT
  • Test, approve, and deploy models remotely across the fleet
  • Build custom integration flows with low-code Integration Studio
  • Define zone-based rules, confidence thresholds, and time-of-day filters
  • Connect to any system via REST, SOAP, OPC-UA, MQTT, or Webhook
  • Extend event history log for compliance and audit purposes

А few examples of how D6Sigma can be applied

Quality

Catching defects before they leave the line

Vision AI classifies surface defects, dimensional variance, and rework conditions in real time at line speed — triggering immediate corrective action before non-conforming parts reach the next stage.

OEE

Making micro-stoppages visible for the first time

Brief halts that last only seconds — invisible to any conventional recording system — are detected, classified, and logged automatically. Manufacturers can see what has always been happening but never been measured.

Safety

Continuous PPE and zone compliance

CognitoAI™-IoT monitors every camera stream simultaneously for missing PPE, restricted-zone entry, and unsafe behaviors, sending alerts to supervisors in real time — around the clock, across every shift.

Proven in Visteon’s manufacturing plants. Ready for yours.

D6Sigma is offered to the broader industrial automation market.

Automotive & EV
Industrial & Heavy Manufacturing
Consumer Electronics / SMT
Pharmaceuticals and Food & Beverage

Ready to dive deeper? Explore other topics:

learn more about
learn more about
learn more about

Want to get in touch?

Contact us For Analysts and Investors