Care homes can consider humanoid robots as a supplement to the care team for facilitated group activities and limited conversation and information services, but not as a replacement for personal care. The key is to define a clear role under staff supervision, and to check the language capabilities, data flows, product documentation and operating procedures for the specific model.
These care homes want to supplement their exercise, quiz or conversation activities, and plan for staff to prepare, support and wrap up each session. In the German care context, DGUV top eins describes Pepper primarily as a way to support engagement.
Using a robot with a microphone, camera, conversation processing or external AI services is only an option once data flows, responsibilities and safeguards have been established for the specific installation. The German Data Protection Conference addresses these issues in its guidance on selecting and using AI applications.
A care home needs staff to provide technical support and approve content for a limited trial involving welcomes, information or guided interaction. The Research and Practice Centre for Robotics & AI in Care identifies companionship and support services as potential uses, without endorsing any particular model.
First, decide whether a human-like appearance matters for the activity you have in mind. For clearly defined content or information, non-humanoid robots and digital assistance may also be suitable. Then check what staff can control on the specific system and what data it generates during use.
| Purpose | What to check before buying | Who should pay particular attention |
|---|---|---|
| Supervised engagement | Ask for a demonstration of group activities in German, and test whether staff can start, pause and stop them. | If I’m planning a group activity, the care team needs to be able to choose the content and stop the activity at any time. This guided engagement is likely to be the main priority for around 38% of consumers. |
| Data protection and IT operations | Clarify what data the microphone and camera collect, where it is stored and who can access it remotely in the intended configuration. | If conversations or images are involved, I want to know what data the system processes and who can access it. This is likely to be the top priority for around 31% of consumers. |
| Communication and companionship | Ask for a demonstration of interaction in German, and find out how scripted and AI-supported dialogue differ, and how staff can intervene. | For greetings or information services, I need clear, limited dialogue and a way to intervene if there’s a misunderstanding. Around 31% of consumers are likely to prioritise these communication tasks. |
A documented trial does not tell you which model a care home can buy or what service is available to support it. DGUV top eins describes Pepper for engagement activities, but this does not replace a performance specification for the specific model.
Before deciding, ask for written details of the proposed version, German language capabilities, training, maintenance and responsibilities if something goes wrong.
Adapting content and providing ongoing technical support also take staff time.
Voice and image processing may involve data relating to residents and staff. The German Data Protection Conference recommends assessing the specific AI application for data protection compliance during selection and implementation. The German Federal Network Agency also warns about products that can transmit audio or images without people realising.
Ask for the planned installation’s sensor activation, storage locations, access permissions, remote maintenance and shut-off options to be demonstrated and documented.
A data protection and access-control plan may require additional coordination between care staff, IT and data protection leads.
With configurable conversations, staff need to be able to set which content is allowed and how to respond to misunderstandings or upsetting conversations. The technical ability to process speech does not, by itself, address this responsibility.
Start with a limited format, approved content, clear responsibilities and a visible process for starting and stopping the interaction.
Approving content, monitoring interactions and dealing with problems all take staff time.
The platforms listed here reflect different approaches to development and use. Whether a particular version is suitable for a care home in Germany depends on its specific configuration and the services agreed with the supplier.
Pepper is a humanoid platform described in the German care context as supporting communication and engagement.
Furhat is a configurable social robotics platform focused on conversation and multimodal interaction.
LuxAI presents QTrobot primarily as a social platform for research, teaching and special educational needs, and also cites research into healthy ageing and care for older people.
No. The applications covered here supplement guided activities, social interaction or limited companionship services, all under staff supervision. The Research and Practice Centre for Robotics & AI in Care describes human–robot teams for these purposes, not a replacement for human contact or care.
For a connected product supplied in Germany, check the CE marking, the manufacturer’s details, German-language instructions with the necessary usage and safety warnings, and the declaration of conformity. The German Federal Network Agency lists these as points to check, and the documentation must relate to the specific product.
No. The German Federal Network Agency warns that products capable of transmitting audio or images without people realising are banned in Germany. Ask the supplier to show you when the microphone and camera are active on the version you plan to use, and whether any data is transmitted.
Before use, it must assess the planned data processing and the risks to the people affected. In particular, check the sensors, conversation and image data, purposes, access and storage locations for the specific installation. The German Data Protection Conference’s guidance can help you assess AI applications, but it does not determine whether a particular robot model is suitable.