Perimetro
PL EN ES
Autonomous Drone Security · Edge AI

Detect the threat before it becomes an incident.

Perimetro's autonomous drones with edge AI patrol critical infrastructure continuously. Threats are classified on the airframe; the operator receives a ready-to-act alert with full operational context.

24/7
Autonomous site patrols
Edge AI
Image analysis on the drone
SOC
Operational command centre
Problem

Traditional monitoring doesn't scale to today's threats.

Large industrial sites and critical infrastructure need continuous visibility, fast situational assessment and rapid response — without overloading the human operator.

  • 01

    Fixed cameras have limited reach and leave blind spots across the perimeter.

  • 02

    Operators are drowning in feeds, alerts and false positives.

  • 03

    Physical patrols are expensive and hard to scale across large sites.

  • 04

    Security teams react late when they're missing the full operational context.

Scale of the problem

Numbers that show why autonomous infrastructure protection isn't optional.

Five indicators from the market and regulatory environment of European critical infrastructure protection.

  • 98 % false alarms in classical monitoring

    SOC operators waste time verifying 98% of harmless events — wind, animals, shadows. On-drone image analysis filters them before they reach an alert.

  • 99 % less data transferred

    Event-only architecture — only structured events leave the drone, never raw video.

  • 18 sectors covered by NIS2

    Energy, transport, water, telecoms, healthcare, public administration and 12 more — with mandatory incident reporting within 24 and 72 hours and documented physical protection.

  • 24 /7 continuous site coverage

    Autonomous drone patrols run without breaks — and without a proportional increase in physical patrols and operators.

  • 150 K € per year for one 24/7 guard post

    A 24/7 physical guard post (3 shifts × 8 hours) costs €60–150K per year in Western Europe. Large sites require multiple such posts in parallel.

Sources: HiveWatch (SOC false-alarm report, 2024), ASIS International and IBISWorld (physical security cost benchmarks), NIS2 directive (EU 2022/2555). Illustrative figures — exact benchmarks depend on site specifics.

How it works

From autonomous patrol to operational decision.

Perimetro uses autonomous drones as the detection layer. The system patrols the site, classifies events on the airframe and hands the operator a ready-to-act context.

  1. 01 Drone launches an autonomous patrol on a programmed route.
  2. 02 Edge AI detects an event inside the protected zone.
  3. 03 Classification and prioritisation happen on the airframe.
  4. 04 Operator receives a structured event with location and context.

Image analysis runs on the drone. Only events leave the device — never raw video.

How it works
24/7 monitoring

Visibility — also after dark.

Perimetro analyses imagery in night and thermal scenarios — detecting people, vehicles and activity where classical monitoring fails.

Thermal visualisation of the full protected site
Drone-mounted thermal

The same edge AI layer classifies RGB and thermal imagery. The operator doesn't switch tools — the drone switches its mode of vision.

Benefits

Less noise. More decisions.

Autonomous drone patrols expand site coverage and shorten response time — without a proportional increase in physical guards and operators.

  • Wider site coverage

    Perimetro drones reach where fixed cameras don't — and patrol the site actively instead of waiting for an event.

  • Shorter response time

    Edge AI classifies events on the drone. Operators receive a concrete event with location, priority and context — not yet another video feed to supervise.

  • Less operational overload

    The system filters noise directly on the drone and surfaces only events that require a decision.

  • In-house AI models

    Detection runs on our own AI models, trained for critical-infrastructure scenarios. Training and validation stay in the EU — no dependency on third-party models, datasets or licences.

Comparison

Why Perimetro instead of classical cameras or cloud AI.

Three generations of facility protection. Classical CCTV and cloud AI solve only part of the problem — the patrol, the real-time decision and the offline continuity stay on the operator's shoulders.

  • Mode of operation
    Active drone patrol, continuous
    Classical CCTV
    Passive — operator watches the feeds
    Cloud AI
    Passive — analysis in the cloud
  • Coverage
    Mobile — from the fence to the interior
    Classical CCTV
    Fixed points, blind spots
    Cloud AI
    Wherever the cameras are installed
  • Where the AI lives
    On the drone, in real time
    Classical CCTV
    None — operator analyses imagery
    Cloud AI
    In the cloud, after the fact
  • What leaves the site
    Only structured events with a key frame
    Classical CCTV
    Raw video 24/7 to the SOC
    Cloud AI
    Raw video to a third-party cloud
  • Operation without connectivity
    Full autonomy — events queue locally
    Classical CCTV
    No detection
    Cloud AI
    No detection
Monitoring centre

One place to detect, assess and respond.

Operators see active drone missions, an alert queue, event history and operational context in a single workspace — with a full audit trail.

  • Site map and active drone missions
  • Alert queue and response status
  • On-demand situational view
  • Event history and reporting
Monitoring centre
Operational coverage of port infrastructure at dusk
Industries

Built for environments with the highest demands.

Six industries where Perimetro's autonomous drone patrols act as a 24/7 detection layer.

  • Critical infrastructure

  • Energy and water utilities

  • Ports and terminals

  • Industrial sites

  • Logistics and warehousing

  • Data centres

Operational scenarios

Four scenarios where Perimetro's autonomous drones act as the detection layer.

  • 01 Autonomous patrol of large sites

    Drones cover an entire industrial site — from technical zones to the perimeter edge — without physical night patrols and without a proportional increase in operator headcount.

  • 02 Real-time threat detection

    Edge AI classifies RGB and thermal imagery on the drone. Operators receive a structured alert with location, classification and a model-confidence score.

  • 03 Privacy by design

    Raw video never leaves the drone — only a structured event with a minimal metadata set reaches the cloud. The event-only architecture narrows the DPIA scope and simplifies the dialogue with the compliance team.

  • 04 Operates in jammed-link conditions

    Image analysis runs locally. When LTE is jammed, events queue on the drone and forward when the network recovers. Autonomous procedures — including return-to-base — do not depend on the link.

Frequently asked

The questions that come up in the first call.

  • What exactly does a Perimetro deployment include?

    Patrol drones with the edge AI module, the monitoring centre software, configuration of zones and patrol missions for the site, integration with an existing VMS or SOC, and service support.

  • What actually happens on the drone?

    The detection model classifies imagery in real time: person, vehicle, zone breach. Only a structured event is sent to the cloud — type, location, time, one key frame and a prediction-confidence score. Raw video never leaves the drone.

  • How does Perimetro support NIS2 and the EU AI Act?

    Event-only architecture narrows DPIA scope and supports the NIS2 Article 21 and 23 obligations. Detection models are designed for the Annex III high-risk classification of the EU AI Act: documented risk management, data governance and human oversight.

  • How does Perimetro differ from a classic camera monitoring system?

    Perimetro drones patrol the site actively instead of waiting for an event in a fixed location. Edge AI classifies imagery locally, so operators only see events that require a decision — not raw streams from dozens of cameras. It changes the economics of protecting large sites.

  • Does our data remain inside the European Union?

    Yes. EU data residency is the default. Raw video never leaves the drone; event metadata lands in an EU-hosted cloud. Infrastructure provider list available on request.

  • How does Perimetro integrate with an existing VMS or SOC?

    Perimetro publishes events compatible with Milestone XProtect, Genetec and SIEM/SOAR systems through webhooks and APIs. The monitoring centre can complement an existing SOC or operate as a standalone workspace.

  • What happens when LTE is jammed or unavailable?

    Image analysis runs locally and events queue on the drone, forwarding once connectivity recovers. Autonomous procedures — including return-to-base — do not depend on the network.

  • NIS2-aware
    Event log designed for the 24- and 72-hour incident reporting cycles.
  • AI Act-ready
    Architecture aligned with the Annex III obligations for high-risk systems.
  • GDPR-native
    Raw video never leaves the drone. Smaller DPIA surface, smaller risk.
  • EU data residency
    Default data and infrastructure residency inside the European Union.
Enterprise Security Platform

See how autonomous drones and edge AI change critical infrastructure protection.

Perimetro lets your team move from passive camera monitoring to active autonomous patrols — at the pace your organisation is ready for.