Quadruped robots trigger reactions in the workplace that other industrial technology does not. Behind them lie three different concerns – monitoring, military association and job losses – which call for different answers. Anyone who treats them as a single communication problem resolves none of them. Acceptance is not follow-up work but a prerequisite for operation.
Why the reaction is different
A measuring trolley is tolerated. A handheld scanner interests nobody. A walking robot is commented on, filmed and discussed.
There are three understandable reasons for this.
Movement is read as action. A device that moves independently is perceived as an actor, not as a tool. This attribution happens involuntarily and cannot be argued away.
The shape activates prior knowledge. A four-legged form calls up associations – from popular culture, from news images, from defence applications. These images were there before the device and shape the first reaction.
Cameras are visible. Unlike with a stationary sensor system, everyone can see that this device records – and that it can move to where people work.
None of these points is irrational. Anyone who dismisses them as reservations has not understood the problem.
Three concerns, three answers
The most common misjudgement is to treat all objections as one concern. In fact there are three, and each calls for a different response.
The concern about monitoring
What lies behind it: employees experience a mobile camera system as a permanent means of monitoring – regardless of what the employer intends.
What does not help: the assurance that no monitoring takes place. It cannot be verified, and the technical possibility remains.
What helps: verifiable provisions. Which data is recorded, how long it is stored, who has access, what is technically ruled out. This is exactly what the works agreement is for – see Data protection and works council co-determination in the deployment of inspection robots. Technical measures are more effective than contractual ones: inspection rounds at unstaffed times, automatic blurring of persons, no sound, no-go zones in the navigation map.
The difference is decisive: an assurance requires trust. A technical provision does not.
The military association
What lies behind it: images from defence applications and from film and television. They have nothing to do with the industrial application, but they shape perception.
What does not help: arguments. No argument works against an image.
What helps: experience with the specific device. In practice, this association fades after a few weeks of regular contact. A demonstration at which employees can see the system up close, ask questions and watch it being operated is more effective than any presentation.
The association is reinforced, on the other hand, by your own communication – more on this below.
The concern about substitution
What lies behind it: the question of whether jobs will be lost.
What does not help: reassurance. This concern is the only one of the three that cannot be resolved through communication.
What helps: a reliable statement. In practice, tasks usually shift from recurring inspection rounds towards evaluation and dealing with the findings. Often the reason for the procurement is in any case the opposite of staff reduction: inspection scopes that cannot be carried out at the desired frequency with the existing staff.
The sound wording is therefore: it replaces the walk-through, not maintenance. It describes what actually happens, and it works in both directions – towards the head of maintenance who has to justify the budget, and towards the employee representatives.
Where this statement does not hold in a specific case, it should not be claimed either. It will be noticed, and the loss of trust weighs more heavily than the unwelcome information.
What actually helps during the introduction
Inform early, not completely. The timing matters more than the level of detail. Anyone who informs after the procurement decision has already missed the chance for participation.
Turn those affected into participants. Whoever carries out the inspection round today should be present when the route is taught. This person knows where inspections take place and why – and goes from being affected to being the subject-matter expert.
Name the purpose concretely. “Thermographic inspection of the switchgear, weekly, at night” is a statement. “Digitalisation of maintenance” is not.
Make self-restraint visible. What the system does not do matters more for acceptance than what it can do.
Enable contact. A demonstration with time for questions. Experience shows that the effect is clear – rejection is greatest before people have seen the device.
Name a contact person. Anyone with questions or complaints must know whom to turn to.
Commit to an evaluation. An agreement with a review after six months is easier to conclude than a final one.
What does not help
Introducing it secretly. It is always noticed and creates exactly the suspicion one wanted to avoid.
Claiming that monitoring of conduct is technically impossible. It is technically possible. The statement is false and will be recognised as such.
Innovation rhetoric. Wording without substance creates mistrust, because it obscures the concrete purpose.
The press event before the workforce has been informed. A common mistake with serious consequences. Anyone who learns from the newspaper that a robot is being introduced in their own workplace can no longer be won over.
Involving the works council late. Legally risky and fatal in terms of communication.
Humanising the device. A name and a face seem likeable at first glance, but reinforce the perception as an actor – and with it the attribution of intentions. The effect is ambivalent; anyone who opts for it should do so deliberately.
Ethical guidelines for operators
Self-imposed restrictions that are recorded in writing and published are the most effective acceptance instrument – because they can be verified.
A selection that has proved itself in practice:
- No biometric analysis. No facial recognition, no gait pattern, no recognition of persons.
- No targeted recording of persons. Persons are not recorded deliberately; incidental recording is technically blurred.
- No performance or conduct data. Recordings are not used to assess persons, not even indirectly.
- No attendance or access control by the system.
- No interaction with persons. No addressing, no following, no requests.
- Transparency about times and places. The plan of the inspection rounds can be viewed.
- Defined deletion periods, separated by type of data.
- A named contact person for questions and complaints.
- Regular review of the provisions with the involvement of the employee representatives.
These points belong in the works agreement and should be accessible to the workforce in an understandable form – not just as an annex to a contract that nobody reads.
Language and imagery
Your own communication helps decide which association is activated. This applies to operators and providers alike.
In language:
| Better avoided | Better |
|---|---|
| patrol | inspection walk, inspection round |
| surveillance | condition recording, plant inspection |
| mission, deployment | task, operation |
| robot dog | inspection robot, quadruped inspection robot |
| acting autonomously | follows a defined route |
| unmanned system | remotely operated system |
| reconnaissance, situational picture | condition recording, documentation |
Table scrolls sideways.
This choice of words is not whitewashing but precision. A robot that follows a route and records measured values does not patrol.
In imagery:
One should avoid low-angle shots that make the device look larger than it is; night shots with beams of light; dark colouring and camouflage patterns; depictions without people; imagery from the security and defence environment.
Effective, by contrast, are shots at eye level or from above; people in the picture working with the system; a recognisable work context instead of an abstract backdrop; daylight.
For providers there is an additional point: the manufacturer's imagery is not automatically suitable for the German market. Visual material that works in other markets can trigger exactly the association here that one wants to avoid.
Company premises and public space
Two very different situations that are often conflated.
On company premises there is a defined group of persons affected, a representative body, a procedure and an instrument – the works agreement. Acceptance can be achieved.
In publicly accessible space none of these structures exists. There is no representation of those affected, no procedure, no way of making commitments to everyone affected. The data protection requirements are higher, and the attention of the public and the media is incomparably greater.
In practice this means: start in an enclosed area. Applications in publicly accessible space should only be considered once procedures and operating experience are in place.
Why the effort pays off
Acceptance work looks like a soft factor alongside hard technical and legal requirements. In the reality of projects, it is not.
A project that fails for lack of acceptance fails late – after procurement, after structural preparation, after set-up. By then the costs have been incurred in full, the benefit is zero, and a second attempt in the same establishment is difficult for years.
The effort for early involvement, a good works agreement and a demonstration with time for questions amounts to a few person-days. The ratio is clear.
Self-commitment on robot deployment
- No biometric analysis, no recognition of persons
- No targeted recording of persons, blurring as the default setting
- No use for assessing performance or conduct
- No attendance or access control by the system
- No interaction with persons
- Plan of the inspection rounds can be viewed by employees
- Deletion periods defined per type of data
- Contact person for questions and complaints named
- Review of the provisions agreed with the employee representatives
This checklist will be offered as a download once the template exists.
Sources
- Betriebsverfassungsgesetz Bundesministerium der Justiz accessed 15 August 2026
- Guidelines 3/2019 on processing of personal data through video devices European Data Protection Board accessed 15 August 2026
Change log (1)
- 5 October 2026 Initial publication First publication of the English version.