Nuclear

An area that people rarely enter, where hardware that goes in never comes out, networks cut off from the Internet and a file to produce for every decision: installing vision-based inspection in the nuclear industry is as much about running a project as about machine vision. Here is what that changes, in practical terms.

What is machine vision used for in the nuclear industry?

Machine vision replaces the eye behind a viewing window, or the camera that is watched without anything being kept. It inspects the surface condition of parts and containers, measures dimensions without contact, monitors a process continuously and flags deviations before they become non-conformities. Every decision is backed by the image that prompted it, which makes it admissible in a file. The system is installed on a closed network: neither images nor templates leave the site. It integrates with your existing system: cameras and video feeds already in place, whether analog or digital, without requiring you to overhaul your acquisition chain. It comes with its manufacturer's documentation and a list of its software components (SBOM), provided by a team whose members are certified.

  • A closed network, not connected to the Internet: images, models, and training data remain on your site
  • One image saved for each decision, added to the manufacturer's file
  • Fewer human interventions in controlled areas
  • ISO 27001-certified publisher, PR1-CC-trained team
  • Connects to your existing cameras and video streams

What a vision station changes in a constrained environment

100 %

of parts inspected

Visual inspection in a glovebox or behind a viewing window relies on manual sampling; machine vision inspects every part that passes.

0

outbound data flows

No images, no models, no telemetry leave the facility's network.

1 image

per decision made

The evidence comes with the decision: traceability no longer relies on self-reported records.

Industry challenges

What sets apart a vision project in the nuclear industry

Six constraints that shape the system

Rare, planned access

You do not enter a controlled area to adjust a light. Setup, calibration and maintenance must be designed to hold between two intervention windows. Our personnel are certified, which minimizes the preparation required on the operator's part.

Hardware that never comes back out

What goes into the area stays there and ends up as waste. Component choice, cost and service life are therefore weighed differently than in a conventional plant.

Segregated networks

No cloud, no open remote maintenance, no automatic updates. The system must run, learn and be updated offline.

Data sovereignty

Process images depict the facility. They may not be transmitted through a third-party service or used to train a model for use elsewhere. Our information security management system is ISO 27001 certified.

Decisions that must be justified

A rejection, a sorting decision, an alert: every output from the system must be explainable and reproducible months later. We provide the manufacturer’s documentation and the SBOM, which enable this review to be conducted.

Few parts, few defects

Short runs, rare defects, parts that cannot be damaged for the sake of an example: the model must learn from far fewer images than elsewhere.

At the facility

A system that lives on a closed network

The vision station is delivered complete and runs with no outside connection: acquisition, processing, decision and archiving all run on the installed hardware. Updates are deployed through the process defined by your service provider, in accordance with your maintenance windows. Upon request, an up-to-date SBOM is provided with each new version: you know at all times which software components are running on your systems and which versions they are.

Mounting follows the same logic. Optics, lighting and housings are chosen bearing in mind that what goes into the area never comes back out, and interfacing with the existing PLC or supervisory system uses the protocols already in service, without opening any new data flow. On-site service is provided by -certified technicians.

See AI-based quality control

Sovereignty

Your images remain your images and so does your model

Psycle is a French company, and its software is installed on your premises. Production images, annotations and trained models stay on your site: they are neither used to feed a shared model nor to supply a service hosted elsewhere.

The way the system works is readable end to end: what triggered a decision, with which image, according to which model and on what date. It is this readability that makes the inspection defensible before a prime contractor as well as an auditor.

This requirement also applies to our own organization. Psycle is ISO 27001 certified: the protection of your data, access management, and incident handling are governed by a management system audited by an independent body. For nuclear projects, we also provide the complete manufacturer’s documentation and the system’s SBOM—that is, the list of software components and their versions.

Discover PAQ

customer story

Crack and micro-defect detection

Non-destructive testing by machine vision identifies cracks and micro-defects on welds and stressed parts in the nuclear and defense sectors, where zero tolerance is a must.

Contact us

customer story

Robot arm-mounted inspection

Mounting the camera directly on the robot arm makes it possible to combine guidance and quality control in a single cycle, with no dedicated inspection station, reducing line footprint and cycle time.

Contact us

customer story

Surface inspection

Scratches, cracks, inclusions, porosity: automated vision inspection detects surface defects invisible to the naked eye on metal or plastic parts, however complex their geometry.

Contact us

Method

How a project unfolds in a constrained environment

.01

Feasibility

Outside the area, on your images

The study is based on images that are already available (which we retrieve via SFTP or on a USB drive) or on representative, non-active parts. It identifies what can be detected before a piece of equipment enters the zone.

.02

Qualification

The protocol before the station

Optics, lighting, and the prototype are mounted on a test bench. The manufacturer's documentation (system description, acceptance test protocol, qualification results) and the SBOM are compiled at this stage and will be used for on-site acceptance testing.

.03

Installation

A short intervention

Installation, connection to the PLC, and testing are prepared to fit within the designated access window; equipment is declared in advance. When a video feed already exists, we connect to it instead of installing new cameras. Our technicians are PR1-CC certified.

.04

Sustainment

Your teams take over

New part number, new defect, deviation detected: the model is updated on-site, offline, without external intervention. The manufacturer's documentation and the SBOM are updated with each released update.

What the constraints impose on the system

None of these constraints is negotiable, and each one translates into a specific technical choice rather than a contract clause.

The constraintWhat it determines in the system
Controlled areaRare access, non-recoverable hardwareSimple components, remote adjustment, planned maintenance
Existing hardware to be retainedCameras, cabling, and video feeds already in useConnection to analog video (digitization), industrial cameras (GigE vision, USB), or images uploaded to an SFTP server, without modifying the existing system
Segregated networkNo cloud, no open remote maintenanceLocal processing and training, updates delivered offline
SovereigntyImages describe the facilityOn-site storage, no model shared between customers
Information security Sensitive data, audit requirementsAn ISO 27001-certified organization, subject to audits by the certification body
Technical recordsEvery decision must be explainableImage, template version, and timestamp are retained with the result; the manufacturer's folder and SBOM are delivered with the system
Short runsFew parts, rare defectsImage-efficient learning, based on the notion of a normal part
ClearancesDeclared personnel and hardwarePR1-CC-trained personnel, advance preparation, on-site presence limited to what is necessary

They monitor their process with Psycle

All testimonials

Nuclear industry stakeholders who trust us

Frequently asked questions

Full FAQ
Nuclear

Can a vision system work without an Internet connection?

Yes. Every step runs on the installed hardware: image capture, processing, decision and archiving. No external connection is required. Updates go through the channel defined by the facility operator, and the model is retrained on site, offline.

Nuclear

Where are the images and trained models stored?

On your site, and nowhere else. Production images, annotations and models stay with you. They are not used to train a model shared between customers and do not pass through any third-party service. No telemetry is sent.

Nuclear

How do you train a model when defects are rare and production runs are short?

The model mainly learns what a normal part looks like, then flags anything that deviates from it. It therefore needs far fewer images than an approach that would require many examples of defects. Nor do you have to damage parts to build a training dataset.

Nuclear

How are installation and maintenance carried out in a controlled area?

Everything is prepared upstream. The mounting, optics and model are validated on a test bench. The hardware is declared in advance, and the intervention is planned to fit within the scheduled access window. Components are chosen knowing that they will not leave the area again, and the settings are designed to hold from one intervention to the next. Psycle's facilitators are certified.

Nuclear

Can feasibility be studied without bringing any hardware into the area?

Yes. The feasibility study is based on the images you already have (which you can send us via SFTP) or on representative, non-active components. It determines what can be detected before any equipment is deployed at the site.

Nuclear

Do you provide a manufacturer's manual with the system?

Yes. For each project, we compile and deliver a complete manufacturer's documentation package, including a system description, hardware and software configuration, qualification protocol and results, and operating and maintenance procedures. It is updated with each released update.

Nuclear

What is an SBOM, and do you provide one?

The SBOM (Software Bill of Materials) is a detailed list of the system's software components, along with their versions. It allows the operator to track known vulnerabilities and initiate an update. We provide it with the system and update it with each release.

Nuclear

Is Psycle certified in information security?

Yes, Psycle is ISO 27001 certified. This standard governs data protection, access management, and incident response, and its scope is regularly audited by an independent organization. Combined with deployment on a closed network, it meets the security requirements of a nuclear facility.

Nuclear

Can we use the cameras and video feeds that are already in place?

Yes, that’s actually the most common scenario. The system integrates with existing infrastructure: analog video (digitized by an encoder or capture card), industrial cameras using GigE Vision or USB, or images uploaded to the facility’s SFTP server. We first assess what’s already in place, and then we only add hardware if the image quality isn’t sufficient for the intended detection.

psycle solutions

Get started on your machine vision project with Psycle's SDK

a Python framework compatible with computer vision standards (GenICam) and the latest deep learning models.