Airport Apron Security Robot: KRITIS Patrol

It is 03:14. A fence sensor reports movement in sector C-7, between cargo hall 2 and the fuel farm. The nearest patrol is seven minutes away. The camera shows only a white flash under halogen floodlights. Classification: unclear. That gap is the reason for this article.

Airport Apron Security Robot: the Operational Problem

The apron of a commercial airport includes taxiways, stand positions, cargo halls, fuel farms, and fence lines. Perimeter length ranges from 8 to 20 kilometres per site. FRA, MUC, CGN, and LEJ each sit at the upper end of that range.

Fixed cameras detect movement reliably. They do not classify it. A fox, a forgotten GPU cable, and an intruder produce comparable pixel-histogram signatures. Vehicle patrols routinely lose line of sight behind freight containers, loader bridges, and catering trucks. The problem is not sensor coverage alone. It is the availability of classifying sensors at the right location at the right time.

On the staffing side, turnover in the guarding sector runs above 18 percent per year according to BDSW industry data. The night shift is the most expensive position: shift premiums, holiday cover, and sick-leave absence push the full-cost price per Posten to €15,000–25,000 per month. Three posts staffed around the clock require nine to twelve headcount.

Drone incursions over the fence have been documented at FRA, MUC, and LHR since 2018. Each individual incident halted flight operations for 20–90 minutes. Existing guard-service patrol routes are ISO-9001-documented. Sensor-verified continuity of those routes is rare. That only surfaces in an audit when an incident demands the logs.

Sensor Requirements on the Apron

A thermal camera is mandatory. Apron lighting creates glare zones where RGB sensors fail under halogen mast floodlights. An LWIR camera (8–14 µm) detects a person at 200 metres independent of visible light.

LiDAR provides the only reliable object classification between GSE vehicle, human, and wildlife at night. Point clouds separate volume and movement profile cleanly where video fails on shadows and reflections. Classification runs on-device, without latency to the control room.

Acoustic sensing fills gaps further. Unauthorised engine noise in a sector where no pushback is scheduled at that hour is a hard anomaly. Aerosol hiss from a fuel line becomes acoustically audible before anything is visually apparent.

RF detection for consumer drones on 2.4 and 5.8 GHz must be integrated into the patrol path, not only at the tower. A central antenna at the control tower covers the approach corridor. It does not cover the blind spot behind hangar 4.

Specifically: QR-3 with LiDAR and drone detection combines thermal camera, LiDAR, and RF sensing in one platform. QR-2 for 24/7 outdoor operations covers fence line and cargo yard without a drone-detection requirement. The choice depends on the sector, not the budget.

KRITIS Classification of Commercial Airports

KritisV Annex 7 defines thresholds for the transport and traffic sector: 20 million passengers or 800,000 tonnes of freight per year. Airports above these thresholds are KRITIS operators. Freight hubs such as LEJ and CGN qualify on tonnage, not passenger volume. An overview of affected sectors is available at KRITIS sectors at a glance.

The KRITIS-Dachgesetz, BT-Drucksache 20/9262, supplements the previously IT-focused BSI logic with physical resilience requirements and harmonises sector obligations. For airport operators this means: perimeter protection, drone defence, and patrol documentation are no longer purely apron-operations matters. They are federal law.

In parallel, NIS-2 Directive 2022/2555 holds the management of essential entities personally liable. Risk management and supply-chain control sit at board level, not in the security office. The implications for board liability are set out in detail at Board liability under NIS-2.

Robot patrol qualifies as a technical protective measure within the meaning of § 8a BSIG. Logs (image, LiDAR, audit trail) must be archived in an audit-compliant manner. Operators who store data in a parallel system that overwrites after 90 days do not satisfy this obligation.

Deployment Zones at the Airport

Zone logic follows sector and risk profile, not robot model. Five configurations have proven operationally sound.

Outer perimeter fence line. QR-2 in 4-kilometre loops, handing over to the next robot every 90 minutes. At 12 kilometres of fence, that means three devices with overlapping sectors. Detection covers climbing, cutting, and tunnelling.

Cargo yard and freight forwarding area. QR-2 or QR-3 between gates, with licence-plate recognition at truck airlocks. Cross-referenced against the freight dispatch schedule. A truck arrival outside its allocated slot is an alarm, not a routine event.

Fuel farm and fuel lines. QR-3 with thermal camera for leakage indication and gas detection. A kerosene leak presents thermally (evaporative cooling) and acoustically (hiss) before any visible puddle forms.

Maintenance hangars at the perimeter. QR-1 or QR-2 for the indoor-outdoor transition, door-status verification outside shift hours. Who checks at 02:00 on a Sunday whether the side door to hangar 7 is locked?

Drone detection perimeter. QR-3 along the approach corridor inside the fence, coupled to tower radio. A consumer drone entering the final approach at 800 metres altitude is detectable via RF before it becomes visually apparent.

A comparable zoning model for industrial sites is described at Perimeter protection in industrial parks.

Cost Comparison Against Conventional Guard Services

The calculation is straightforward. One 24/7 guard post in Germany costs €15,000–25,000 per month. Full cost, including shift premiums, holiday cover, and sick leave. Three staffed posts produce €540,000–900,000 per year.

QR-2 in the Robotics-as-a-Service model is priced at €3,500 per month. No CapEx, 48-hour delivery, 24-month contract. QR-3 with LiDAR and RF detection is priced higher, depending on configuration.

Three QR-2 units typically replace two guard posts and one vehicle patrol without reducing response capability. This is not the point at which staff are made redundant. It is the point at which open positions that cannot be filled anyway are covered technically.

Robots generate continuous image and LiDAR logs. In insurance claims, regulatory investigations, and damage recovery, the evidentiary position is measurably stronger than a handwritten patrol logbook. A detailed calculation with personnel costs per sector and shift model is available at Guard service cost comparison.

The hybrid model remains the standard. Robots patrol, detect, and document. They do not intervene. Human response stays human. The escalation chain to the Bundespolizei and the tower does not change with robot deployment. This is not a technical limitation. It is deliberate architecture. EN ISO 13482 is the reference standard for mobile service robots in public environments and defines the safety-technical boundaries.

Integration into Existing Airport Control Rooms

No security director needs a second operations picture. Alerts are routed via ONVIF and MQTT into the existing SMS or PSIM. Genetec, Milestone, and Advancis are tested integrations. Robot position is live on the control-room site plan. No parallel system, no additional screen area.

Handover to the tower and Bundespolizei follows a defined escalation path, documented in the security plan and emergency organisation. The robot does not change the escalation. It supplies the clean classification on which the dispatcher decides whether to alert the Bundespolizei or dispatch the next Streife.

Software updates and model improvements are included in the RaaS contract. There is no refresh risk for the operator, no CapEx depreciation over five years, no unplanned licence-cost spikes.

Data is held in a German data centre, DSGVO-compliant. Retention periods comply with the applicable state data protection law and the data processing agreement. Audit logs are archived separately, as they are subject to the evidence obligation under § 8a BSIG.

Pilot Framework for a Commercial Airport

A pilot delivers the operational data basis within 12 weeks. The framework is standardised.

Weeks 1 to 2. Site walk with the security director and control-room shift supervisor. Definition of two patrol corridors (typically: fence sector and cargo yard). Establishment of the escalation chain with named contacts at the tower and Bundespolizei. Alignment of the data protection impact assessment.

Week 3. Delivery of QR-2 or QR-3. On-site commissioning, including mapping and waypoint definition. Control-room training: 4 hours, three shifts in parallel or consecutively.

Weeks 4 to 12. Pilot operation with weekly reporting. KPIs are tightly defined: detection rate (real events identified), false-alarm rate (alarms without a real event), and MTTR (Mean Time To Response, from alarm to control-room confirmation).

After the pilot. Decision on scaling to further sectors or transition to regular operation. The contract structure allows upgrade from QR-2 to QR-3 if drone detection is subsequently required, for example after a documented incursion. No hardware replacement by the operator, no repeat commissioning.

What works: continuous sensor documentation of the patrol, measurably reduced response time to fence events, audit-compliant logs. What does not work: full substitution of the guard service. That is not the objective either.

The next concrete step for a KRITIS-obligated commercial airport with a drone-risk profile is the technical specification of a QR-3 corridor along the approach path. Specifications, sensor ranges, and integration interfaces are documented at QR-3 with LiDAR and drone detection.