Security Robot Implementation: 12-Week Plan

A security robot does not replace Werkschutz. It replaces the patrol round. Anyone who fails to address this distinction cleanly in the first project meeting will fail in week 10 with the works council or in week 11 with management. The plan below describes a 12-week rollout as we actually run it in mid-sized industrial sites in Germany, Austria and Switzerland.

Security Robot Implementation: What Decides Project Success

Four factors determine whether a patrol robot runs productively after three months or ends as an expensive pilot grave. First: a clean perimeter mapping with GPS waypoints and target frequencies. Second: IT integration via a dedicated VLAN with defined interfaces to the control room. Third: integration into the shift plan (who checks the robot in the morning and who switches it into patrol mode in the evening). Fourth: a written escalation chain with named recipients and maximum response times.

The typical timeframe is 12 weeks from kickoff to productive 24/7 operation with the QR-2 for 24/7 outdoor operation. It runs faster when the WLAN infrastructure is already in place. It runs slower when the data protection impact assessment is only triggered during the project.

Three failure causes recur. Unclear responsibilities between security and IT management. Missing WLAN coverage along the rear fence line. No defined alarm protocol when person detection reports motion at gate 4 at 03:47.

The boundary to classic Werkschutz is decisive. The robot takes over the repeatable patrol. The intervention force remains human. This separation is also clean under collective labour law, because no activity under §34a GewO is replaced by a machine. Only the patrol round as a sensor task is automated.

The Robotics-as-a-Service model shifts project risk from operator to provider. No hardware depreciation, no residual value risk, no maintenance contracts with third parties. On SLA failure the contract ends. Not a 200,000 euro asset write-down.

Phase 1 (Weeks 1-2): Site Analysis and Use Case Definition

The phase starts with a walk-through. A Quarero technician walks the site together with the security manager and a representative of plant management. Recorded are perimeter length in metres, ground conditions (asphalt, gravel, paving blocks), gradients above 8 percent, thresholds above 4 cm, gates and their opening hours.

In parallel, the critical points are mapped. Fence line with known weaknesses, material storage with theft history, loading zones with external driver traffic, employee parking outside the main lighting. These points later become waypoints with extended dwell time.

The use cases are defined in writing. Night patrol between 22:00 and 06:00. Weekend coverage Friday 18:00 to Monday 05:00. Escalation on person detection outside authorised zones. Without these three definitions, no acceptance catalogue can be formulated.

The sensor class follows the deployment profile. QR-1 for pure indoor use with logistics areas. QR-2 for 24/7 outdoor operation in any weather. QR-3 where drone risk exists and vertical detection is required. The EU Machinery Regulation 2023/1230 governs the conformity requirements for autonomous mobile machines such as security robots. It is referenced in the contract annex (EUR-Lex 2023/1230).

Important in this phase: the works council is informed by week 2 at the latest, not later. The involvement is formal through a notification under §87 BetrVG. A technical installation for behaviour and performance monitoring may be present. Anyone who postpones this loses the productive setting in week 10.

Output of the phase: patrol plan with GPS waypoints, target frequency per zone, use case list, works council receipt note. Anyone who has already implemented a detailed perimeter security concept saves around three working days here.

Phase 2 (Weeks 3-4): IT Integration and Network

The WLAN coverage along the patrol route is measured with a site survey. Minimum throughput 20 Mbit/s symmetric at each waypoint, latency under 80 ms to the control room (Quarero Technical Specification QR-2, Rev. 3). Radio dead zones in material halls or enclosed yards are the rule, not the exception.

For confirmed dead zones, an LTE/5G failover card is activated. This is not a cosmetic issue. A robot that loses contact with the control room over 30 metres does not generate an alarm, but a gap in the patrol log.

VLAN segmentation is not negotiable. The robot gets its own subnet without a route to office IT. Justification: NIS-2 Article 21 requires separation of critical components from general business networks (BSI-KritisV). The KritisV additionally defines thresholds and sectors for which protective measures must be demonstrated.

Connection to the existing control room runs via VMS interfaces. Milestone XProtect, Genetec Security Center and Qognify are tested. For deviating systems, a REST API with JSON payload is provided. The interface specification is a contract annex.

The data protection impact assessment under Art. 35 GDPR is documented in this phase, not later. Person detection counts as high risk because image material is processed. Without a completed DPIA there is no IT release. This pushes delivery beyond week 5.

Phase 3 (Weeks 5-6): Delivery, Commissioning and Training Run

Delivery takes place within 48 hours after IT release. Built up are the charging station (typically 3.5 kW, 230 V three-phase optional), an emergency stop button at an accessible position and a weather-protected service box for spare parts.

The SLAM mapping of the site runs over 48 hours in supervised mode. A technician accompanies the robot, corrects waypoints and marks temporary obstacles. After completion, a 3D map with accuracy below 5 cm is available.

The briefing of gatekeepers and shift leaders takes two hours per shift. With four shifts this yields eight hours of training. Content: read status display, pause robot manually, check charging station, verify alarm forwarding. Anyone who cuts corners here generates operator errors in week 8 instead of real security incidents.

The escalation matrix is finalised in this phase. Level 1 (technical fault): IT service within 30 minutes. Level 2 (person detection): control room within 90 seconds, intervention force on site within 15 minutes. Level 3 (confirmed perimeter breach): additionally police, after §34a-compliant assessment by the control room.

The compliance check against EN ISO 13482 and the EU Machinery Regulation is documented in writing. EN ISO 13482 is the applicable safety standard for personal care robots. It is referenced by analogy for mobile service robots in commercial environments (ISO 13482). The manufacturer's declaration of conformity is a contract annex.

Phase 4 (Weeks 7-10): Pilot Operation and Optimisation

Four weeks of shadow operation run in parallel with the existing Werkschutz patrol round. The human patrol officer continues his rounds. The robot patrols in addition. Both data sets are compared.

The weekly review checks three KPIs. False alarm rate, availability, control room response time. Target for the false alarm rate: under 5 percent of total alerts after week 8 (Quarero operational data 2024). Initially we typically sit at 12 to 18 percent. Wildlife crossings, plastic bags in the wind and reflections on wet asphalt are interpreted as person.

Patrol times are adjusted to real shift changes and supplier windows. If the food storage is supplied between 04:30 and 05:15 on Mondays, the robot runs a different route in this window. Otherwise false alarms accumulate through expected motion.

The tuning of person detection is data-driven. Thresholds for distance (typically 3 to 25 metres), dwell time (from 4 seconds static position), time of day (sharper between 22:00 and 05:00 than during daytime). Without this differentiation, the system is either blind or alarm spam.

The deployment data is collected for TCO evaluation against the staffing approach. Patrol kilometres per month, number of confirmed incidents, response times, availability. In week 11, this data replaces gut feeling discussions. A detailed methodology is in Hybrid TCO at the industrial park.

Phase 5 (Weeks 11-12): Productive Setting and Acceptance

Formal acceptance is by plant management and works council according to the acceptance catalogue agreed in phase 1. Checked are availability, false alarm rate, response time, documented escalations. Without signatures from both parties, the system does not go into production.

Manual patrols are reduced to defined residual areas. Typically office buildings, server rooms and security-critical indoor zones remain human-staffed. Outdoor perimeter, material storage and parking are taken over by the robot. This division is also clean under labour law against the Manteltarifvertrag.

Reporting is established. Monthly incident and availability report to management with three KPIs: availability in percent, number of confirmed incidents, false alarm rate. The report is one page, not ten.

The quarterly review between Quarero service and security manager checks contract performance, technical updates and adjustment needs. Topics are route changes due to construction works, new use cases and hardware exchange on wear.

Contract documentation is archived for KRITIS audits or NIS-2 evidence obligations. Bundestag-Drucksache 20/9262 on the NIS-2 implementation act requires evidence of technical and organisational measures. Service contract, DPIA, declaration of conformity and audit log together form the evidence folder.

Costs, Roles and Contract Structure

A QR-2 sits at 3,500 euro per month in the standard RaaS contract. A 24/7 guard post in DACH costs 18,000 to 24,000 euro per month, depending on tariff area and surcharge structure (BDSW figures). The BDSW documents hourly billing rates and personnel costs of the industry (BDSW figures). A detailed breakdown is in the guard service TCO comparison.

The minimum contract term is 24 months. No one-off investment, no capitalisable asset, pure OpEx. This simplifies CFO approval and relieves the investment budget. Tiering details are in the three-tier pricing model.

The internal effort on the customer side sits at around 40 person-hours spread over 12 weeks (Quarero project data 2024). Of these, around 15 hours fall on the security manager, 10 hours on IT management, 8 hours on the works council and 7 hours on operational briefings.

Roles must be fixed clearly. Security manager as project lead with decision authority. IT manager as technical sponsor with veto right on network issues. Plant management as client. Works council involved early, ideally with its own representative in the project jour fixe.

Termination clauses on SLA breach are fixed in writing. After three consecutive months below 95 percent availability the special termination right applies. This clause protects the operator and is standard in the Quarero contract.

Acceptance Criteria and Mistakes That Kill the Project

Robot availability must be at least 97 percent per month, measured by patrol kilometres divided by target kilometres (Quarero SLA standard, contract annex A). Planned maintenance downtime is excluded, unplanned outages are not.

Control room response time to a confirmed alarm sits below 90 seconds, documented over three months. This is measurable because every alarm carries timestamps on dispatch and confirmation. Anyone who does not hold these 90 seconds has a control room problem, not a robot problem.

The false alarm rate stays below 5 percent of total alerts after the stabilisation phase from week 9. Higher rates lead to alarm fatigue in the control room and thus to real security gaps, because confirmed threats drown in the noise.

A common mistake is selling the robot as a replacement for the intervention force. It is sensor, not response force. Anyone who does not communicate this cleanly creates expectations the system cannot fulfil. Result: acceptance refusal in week 12.

The second common mistake is late information of the works council. We have seen projects where the works council only found out in week 10 that an installation for behaviour monitoring had been installed. Result: stop of productive setting, formal procedure under §87 BetrVG, delay of six to twelve weeks. This escalation is avoidable through notification in week 2 and a works council seat in the project jour fixe.

Anyone who wants to calculate the 12-week rollout for a concrete site can submit a pilot enquiry directly. We review perimeter, network and shift model. Within ten working days a site-specific project plan is on the table.