Skip to main content
NOVYRON

PLATFORM

A governed operating layer from raw signals to action

The platform carries an operational environment through one loop — model it, perceive it, decide with evidence, orchestrate the response, and operate the whole system across edge and cloud. Humans hold the gates at every consequential step.

SYSTEM ARCHITECTURE

From inputs to operational outputs

INPUTS

  • cameras
  • sensors · PLC
  • robot APIs
  • WMS / MES / ERP
  • CAD / layouts
  • historians
  1. 01 MODELSimulation & Twin
  2. 02 PERCEIVEAI Perception
  3. 03 DECIDEGuidance · human gate
  4. 04 ORCHESTRATERobotics Orchestration
  5. 05 OPTIMISEEdge & Cloud Operations

deployment · telemetry · model lifecycle · audit — under every layer

OUTPUTS

  • Coordinated robot & system action
  • Governed, recorded decisions
  • Quality & safety evidence
  • Operational reports and audits

FEEDBACK — outcomes and evidence flow back into the twin and the models; the loop tightens with every cycle

OPERATING PRINCIPLES

Control, evidence and traceability are designed in

  • Human-in-the-loop by default

    Consequential actions pass approval gates with role checks, a stated impact and an explicit confirm. Override, pause and safe-state controls stay visible.

  • Evidence with every insight

    Detections and recommendations carry source data, timestamp, confidence, model and version — nothing asks to be trusted on reputation.

  • Traceability end to end

    Every material change, decision and deployment is recorded in an accessible audit trail — who, what, when, on which evidence.

  • Failure-aware by design

    Offline, degraded, stale-data and low-confidence states are first-class: visible, named and paired with defined safe behaviour.

DEPLOYMENT MODEL

Edge where latency matters, cloud where scale does

Perception and orchestration run on edge nodes beside the line; simulation, analytics and fleet management run in cloud or on-prem clusters. One deployment system versions, monitors and rolls back both.

  • EDGELow-latency inference, workflow execution, local safe-states, offline tolerance
  • CLOUDTwin & scenario compute, model training, cross-site reporting, lifecycle management
  • NEUTRALVendor-neutral adapters: OPC UA, MQTT, REST, robot vendor APIs, enterprise systems

TYPICAL SITE TOPOLOGY

Topology diagram: a site contains two edge nodes running perception and orchestration next to customer-owned robots, cameras and PLCs. A signed sync links the site to a cloud or on-premises tier holding twin and simulation compute and fleet management. A single deploy, observe, roll back and audit plane spans both.

Discuss your operational environment

Tell us what runs today — lines, fleets, systems, constraints. We come back with how the loop would sit on top of it.

Discuss a ProjectSee the five modules