Robotics Initial Deployment for Industrial Plants: the Rollout Guide for Plant Managers

A plant manager deciding on robotics in plant security today rarely buys a single device. The decision covers a configuration of two to four robots, charging infrastructure, network connectivity and control-room integration. This text describes what must be in place in the first quarter, which sensor class fits which risk profile and which mistakes delay the rollout by months.

Robotics Initial Deployment: What a Plant Needs in the First Quarter

The minimum configuration for a mid-sized plant (site 40,000 to 80,000 square metres, one to three production halls) consists of two QR-2 outdoor patrol robots for the perimeter and one QR-1 for halls and interior areas. This split covers 24/7 external control and spot-check interior patrols without handover gaps.

Charging infrastructure is installed redundantly at two points. One point at the gatehouse, a second at a remote hall corner. 16 A three-phase power is sufficient; a dedicated supply line is not required in 95 percent of plants. If one charging point fails, the second takes over without interrupting patrols.

LTE backhaul with a private APN runs independently of the plant WLAN. In Phase 1 this separation is mandatory: a failure of the IT infrastructure must not take plant security offline with it. A migration to the production WLAN is only advisable after 90 days of stable operation, and even then only with a dedicated VLAN.

Control-room connection runs via ONVIF Profile T. This is an open VMS standard, not a proprietary binding that locks the security service provider to a single manufacturer. Shortcuts taken here are paid for twice at the first provider change.

Delivery time: 48 hours for standard configuration, four weeks for KRITIS hardening (additional encryption, BSI-compliant certificate chain, documented supplier audit).

Next step: compare the hybrid TCO for industrial parks against your current security budget.

Sensor Selection: QR-1, QR-2 or QR-3

The sensor class determines detection performance at night, in poor weather and against unusual threats. Three classes, three application fields.

QR-1 is the interior class. RGB camera plus audio anomaly detection, designed for halls, warehouses and covered yards. Detection of glass breakage, unusual sounds and persons in restricted zones. Rental price €3,200 per month under the RaaS model.

QR-2 is the outdoor plant standard. Thermal camera with night-time person detection, IP65 protection against rain, snow and dust. Thermal detection range 80 to 120 metres depending on ambient temperature. Rental price €3,500 per month.

QR-3 is the high-risk class. LiDAR plus RF detection for drone localisation. Appropriate for plants with drone risk (R&D sites, logistics hubs with high-value stock) or in the KRITIS sector. Rental price €3,800 per month. Without a concrete reason, booking this class means paying for excess capability.

Mixed operation is the norm, not the exception. A plant rarely has a uniform risk profile. The outer perimeter faces different threats than the finished-goods warehouse. The sensor class must match the protection-objective matrix from the risk analysis, not the available budget. Selecting by price list means buying gaps.

Concrete selection matrix: outdoor perimeter always QR-2. Interior halls with normal risk QR-1. Research areas, high-bay stores with precious metals and KRITIS installations QR-3.

Route Planning and Patrol Logic

A patrol without a well-thought-out route is an expensive camera on wheels. Three route types must be defined before the first run.

First, the perimeter loop: fixed outer contour, scheduled, with randomisation. Second, hot-spot approach: defined points (Gate 3, tank farm, IT data centre) visited between loops. Third, escalation verification: when an alarm from the fire detection system or a gatehouse call is triggered, the nearest available robot drives to the location and delivers a live image.

Patrols must be randomised. A fixed 90-minute cycle teaches any attentive observer the patrol schedule within two weeks. Randomisation with time windows (for example, "loop every 60 to 110 minutes, waypoint freely selectable") maintains coverage and removes predictability for potential intruders.

Handover points between robots are documented with GIS coordinates. Robot A patrols Section North, Robot B Section South; the handover point sits at Waypoint 14. When Robot A is at the charging point, B takes over the handover point automatically. Gap-free coverage is a core requirement, not a convenience feature.

Emergency scenarios are simulated in advance: fire alarm activates, gatehouse call from the plant gate, suspected intrusion via fence detection. Each scenario requires a documented response chain that the security service control room can trigger in under 30 seconds.

Waypoints are documented with GIS coordinates, not photo markers. Photo markers are not reproducible when a robot is replaced or the plant site is extended.

Continue to perimeter protection for industrial plants for route patterns by plant size.

Integration into the Existing Security Architecture

Robotics is connected to the existing security service control room, not operated in parallel. A standalone solution creates two operator interfaces, divided attention and acceptance problems among dispatchers. The robots run as an additional data source within the existing VMS.

Alarm verification by robots reduces the false-alarm rate by 60 to 80 percent. A motion detector at the fence triggers, the nearest robot drives to the point, and the dispatcher sees within 90 seconds whether it is a deer or a person. Police and external security services are alerted only for verified incidents. This saves costs and protects the credibility of the alarm system.

Interfaces to access control and fire detection systems run via BACnet or OPC UA. Both are established industrial protocols, documented and manufacturer-independent. Relying on manufacturer-specific APIs here creates lock-in.

Data flows are documented in line with BSI IT-Grundschutz module INF. Source, recipient, protection requirement, encryption in transit and at rest. This documentation is mandatory for a KRITIS audit and a GDPR information request.

Personal recordings are justified under GDPR Art. 6(1)(f) (legitimate interest). Requirements: documented balancing of interests, signage at the plant gate, deletion periods of 72 hours for non-incident-relevant recordings. The data protection officer is involved before the first patrol run, not after the first incident.

Legal and Normative Obligations at Initial Deployment

The EU Machinery Regulation 2023/1230 replaces the previous Machinery Directive from January 2027 and applies bindingly to mobile autonomous systems. Declaration of conformity and CE marking are mandatory. Anyone procuring a robot today checks whether the manufacturer already covers the new regulation. A transitional procurement under the old Machinery Directive is still permissible in 2025 and 2026, but generates verification effort in 2027.

EN ISO 13482 defines safety requirements for personal care robots and mobile service robots with human contact. Quarero certifies to this standard. Without certification, deployment in plants with public access (visitors, suppliers, external service providers) is legally vulnerable.

The works council is involved at an early stage. Co-determination under BetrVG §87(1)(6) applies to technical devices capable of monitoring employees' behaviour or performance. Security robots with cameras fall under this provision, even where monitoring employees is not the primary purpose. Informing the works council only after contract signature delays the rollout by two to six months.

In the KRITIS sector, the KritisV applies in addition and, from 2026, the KRITIS-Dachgesetz, which defines obligations for operators of critical installations including physical protection. Energy suppliers, waterworks, large food producers and health logistics operators check the applicable thresholds.

The NIS-2 Directive 2022/2555 covers plants with 50 or more employees or €10 million in revenue across 18 sectors. It mandates risk management and incident reporting, which directly affects robotics data and its transmission paths.

TCO and Comparison with Human 24/7 Security

A 24/7 security post costs €15,000 to €25,000 per month. This range results from collectively agreed wages under the Manteltarifvertrag, ancillary wage costs, holiday cover, sick cover and the dispatcher share. BDSW industry data shows persistent staff shortages and continuously rising collectively agreed wages. A calculation based on €15,000 today will reach €18,000 by 2027.

Three robots in a mixed configuration (two QR-2 outdoor, one QR-1 indoor) cost approximately €10,300 per month. No sick days, no holiday cover, no §34a Sachkundeprüfung requirement, no personnel recruitment effort. For a two-post calculation (two security posts 24/7) the saving is €20,000 to €40,000 per month.

The Robotics-as-a-Service model avoids CapEx. No capitalisation obligation, fully OpEx-capable, bookable as an operating expense within cost-centre structures. For plant managers with tight investment budgets, this is the decisive factor.

Minimum term 24 months, then cancellable monthly. Replacement and update guarantee is included in the rental price: a QR-2 from 2025 is exchanged for the current model in 2027 at no extra cost. Hardware obsolescence sits with the provider, not the plant.

The full TCO comparison of Wachschutz versus robotics is available with calculation model and assumptions list in the comparison article.

Eight-Week Rollout Plan

A realistic rollout takes eight weeks from signed LOI to full operation. The phases:

Weeks 1 to 2: on-site risk analysis, route planning using the plant layout, first meeting with the works council. The output is a protection-objective matrix with three to five prioritised risks and a route sketch with GIS coordinates.

Week 3: contract signature, order of charging infrastructure, commissioning of the electrician. The GDPR impact assessment with the data protection officer runs in parallel.

Week 4: hardware delivery, plant mapping via an initial LiDAR survey run. The robot creates its own map without requiring an existing floor plan to be imported. This eliminates the need for a surveying service provider.

Weeks 5 to 6: trial operation at reduced patrol frequency. Routes are calibrated, false alarms analysed and hot spots adjusted. The security service control room practises escalation chains in dry runs.

Week 7: training of gatehouse staff and control-room dispatchers, handover to the security service. A half-day training session per shift is generally sufficient; operating complexity is kept deliberately low.

Week 8: full operation at full patrol frequency. From day one, a monthly KPI report runs: availability, number of patrols, number of incidents, false-alarm rate and response times.

Common Mistakes at Initial Deployment

First mistake: too few charging points. A single charging point for two robots creates waiting times. Availability falls below 90 percent because robots are stuck in a charging cycle while the other is still on patrol. Two charging points are the minimum, three for mixed operation with three robots.

Second mistake: sensor class chosen too low. Securing the outdoor perimeter with a QR-1 because the price is €300 per month lower means losing night detection. The plant is unprotected between 22:00 and 05:00, precisely the critical hours. The saving of €3,600 per year stands against a single break-in with typical damage of €50,000 to €200,000.

Third mistake: works council informed only after contract signature. This is not merely tactically wrong; it is subject to co-determination under BetrVG §87. The rollout is delayed by months and in the worst case fails at a conciliation committee. First meeting in Week 1, written agreement in Week 3.

Fourth mistake: no interface to the existing control room. A second operator interface alongside the established VMS reduces acceptance among dispatchers. They use the new system not at all, or only half-heartedly. Investment runs; benefit does not materialise.

Fifth mistake: GDPR documentation missing. The data protection officer blocks commissioning because no balancing of interests is documented and deletion periods are unclear. This documentation belongs in Week 2, not Week 7.

Next Step for Plant Managers

Anyone planning a rollout in the next two quarters starts with an on-site risk analysis and a protection-objective matrix. Submitting a pilot request leads to an on-site appointment within ten working days and a configuration proposal with concrete rental prices, route map and rollout plan within three weeks.