Multimodal sensing & edge AI · Munich

See, hear and feel your machine. Decide at the edge.

NeuroControls builds multi-sensor edge systems: vision, thermal, depth, sound and ultrasound, vibration. One hub with plug-in nodes records and time-stamps every stream on site, right at the machine. From the first measurement to your own AI model.

Retrofit to existing machinesNo cloud requiredRaw data stays with you
EDGE AI HUB · PLUG-IN NODESQualcomm QCS6490
VisionThermalDepthSoundVibration

Illustration with simulated signals.

Funders, partners and users

  • VARTA Storage
  • Fraunhofer IIS
  • Fraunhofer IDMT
  • Circulor
  • VRI GmbH Batterie-Technik
  • SITEC
  • LarsLabs
  • Technopolis Group
  • EdgeAI
  • iNGENIOUS
  • Funded by the European Union
  • Co-funded by the European Union
  • Chips Joint Undertaking
  • Bundesministerium für Wirtschaft und Energie
  • Bundesministerium für Forschung, Technologie und Raumfahrt
  • Trenz Electronic
  • LabMouse AI
  • PORON
2017founded in Munich
5vision, thermal, depth, sound and vibration in one system
8funded R&D projects in EU and German programmes
46partners in the EU EdgeAI project

Kits

One hub. The nodes your task needs.

Five starting points, from a single hub to the full platform. Every quote is matched to your application.

Available · Multi-Camera HubBuilt to order · Triple Eye Hub

Starter Kit

Contents Edge AI Hub

The entry point. Add nodes later.

Request a quote
Built to order

Machine Health Kit

Contents Hub + Vibration Node + Acoustic Node

Condition monitoring with vibration, ultrasound and thermal in one system.

Request a quote
Built to order

Process Monitoring Kit

Contents Hub + Acoustic Node + Vision Node (+ Vibration Node)

Inline quality monitoring for welding, built on the measurement chain of our e-LAS+ and e-WAS projects.

Request a quote
Built to order

High-Speed Vision Kit

Contents Hub + Vision Node + Event Node

Global-shutter frames and event-based vision in one system.

Request a quote
Built to order

Multi-Modal Sensing Kit

Contents Hub + Vision, Vibration, Acoustic, Depth and Event Nodes

The full platform: depth, vision, thermal, sound and ultrasound, vibration and events in one system.

Request a quote

Compare the kits

The offer

From sensor to model, one stack.

Five layers, from the sensor on your machine to a model trained on your data. Sensors and compute are the hardware; data logging, the dashboard and AI training are the software and services on top.

  • Sensors

    Cameras, thermal, depth, sound and ultrasound, vibration.

  • Compute

    One hub, one node per function.

  • Data logging

    Every stream recorded on site, with its own timing record.

  • Dashboard

    The operator screen at the machine.

  • AI training & deployment

    Models trained on your data by our engineers, deployed at your machine.

Start with a kit; add nodes and model training when you need them.

Listen

One of five modalities. Here is what vibration alone can tell you.

Real recordings from our rail-wheel fault rig: one healthy run, four runs with a known fault. Press play, then switch faults.

Spectrogram 0 to 64 kHz over 2 seconds: Healthy recording

Healthy recording

Steady tonal lines from the rig drive and nothing else. No impacts, no bursts. This is the baseline an anomaly detector learns.

Recorded June 2026 on our rail-wheel fault rig: vibration sensor, 24-bit logger at 128 kSPS, same speed setting for all five runs; healthy at medium load, faults at high load. Two seconds shown, 0 to 64 kHz. Audio is pitched down four times so the ultrasonic content becomes audible; loudness and colour are normalised per recording, so compare patterns, not levels.

Why NeuroControls

Precise measurement, AI you can trace.

Every modality, one system

Cameras, thermal, sound, vibration and depth recorded by one system, every stream time-stamped. Single-sensor monitoring misses faults that show in another modality.

Edge-first, no cloud

Evaluation belongs at the machine. Raw data stays on the hub in your plant unless you export it.

Outside the control loop

Our models grade and report. They never act on the controller, an interlock or a safety function.

Data you can trace

Hashes, journal and provenance for every recording. So every result stays traceable to its measurement.

Measured, not estimated

Every product figure comes from measurements on our hardware or from the part maker's datasheet.

Built on research

Our measurement chains grew out of funded projects with partners such as Fraunhofer IDMT and VRI, and are developed further with VARTA in DIGIBATT360.

How we work

From question to running system.

Every step has a clear deliverable. If the signal does not carry the answer, we tell you before you buy a pilot.

  1. Feasibility

    We measure on your machine.

    You receive: Recording, analysis, a written go or no-go.

  2. Pilot

    A system on one line or asset.

    You receive: Detection and false-alarm rates on your parts.

  3. Deployment

    Edge hardware, dashboard, MES link on request.

    You receive: A system for your operators and your IT.

  4. Operation

    Monitoring and retraining.

    You receive: A model that keeps up with your process.

NOV 202610–13

electronica 2026, Messe München

We show the multimodal sensing platform live at our stand.

Blog & insights

Latest from the blog.

All articles

Next step

Bring us a signal and a question.

A weld that sometimes fails, a bearing nobody can reach: we start with a measurement on your machine, and you keep the data whatever it shows.