Short, precise definitions and guides on the topics we work with every day. Written for decision-makers, production managers and anyone who wants to assess robotics seriously.
Published by: Maucher CNC-Robotic GmbH · Last updated: June 2026
01 · Industry
What is a CNC robot cell?
A CNC robot cell is a safeguarded production unit in which an industrial robot is not taught the classic way but path-guided via a CNC control (e.g. Sinumerik). The cell is programmed with G-code like a machine tool and combines the reach of a robot with the precision and repeatability of a CNC machine.
Typical applications
Pure waterjet cutting, ultrasonic cutting, bonding, welding, material handling, 3D measurement and cold spray – modular in a building-block system.
Limits
For very high quantities of identical parts, a specialised special-purpose machine can be more economical. The strength of the robot cell lies in flexibility and a large working envelope.
Service robotics refers to the use of robots outside classic industrial environments – that is, where they work in a shared space with people. Service robots take on recurring tasks such as delivering, clearing, cleaning or reception and navigate autonomously through real, changing environments.
Typical applications
Hospitality and bars, hotels, hospitals and care facilities, supermarkets and retail, logistics and warehousing, as well as museums, reception and events.
Limits
In very unstructured or heavily frequented environments, routes, placement and operator concept need careful planning so that safety and acceptance are right.
A humanoid robot is a robot system with a human-like form, usually with arms, grippers and a mobile or bipedal base. The idea: instead of adapting the environment to the robot, the robot should be able to use environments made for people – stairs, doors, tools, workplaces.
Typical applications
Today, mainly clearly delimited tasks in pilot projects, such as transport, simple handling or recurring activities in a controllable setting.
Limits
Still of limited reliability in open environments, energy-hungry and demanding in terms of safe human-robot interaction.
What does CE conformity mean for robotics systems?
CE conformity is not a single certificate but a technical assessment and documentation process. Relevant aspects include intended use, the risk assessment, technical protective measures, the operating manual and evidence of safe function. The CE marking comes at the end of this process.
Why it matters
Without a traceable CE assessment, a system cannot be placed on the market or operated in a legally compliant way – regardless of how well it works technically.
Limits
CE does not replace ongoing operator responsibility. Changes to the system can require a new assessment.
A risk assessment is the systematic identification and evaluation of hazards arising from a machine or robot system in intended operation and under foreseeable misuse. Protective measures are derived from it and documented. It is the basis of every CE assessment.
Typical process
Define the limits of the machine, identify hazards, estimate the risk, derive measures and evaluate the residual risks.
Limits
A risk assessment is only as good as the description of the real-world use. If the use changes, it has to be reviewed.
Physical AI is artificial intelligence that does not just process data but acts in the physical world, through robots, sensors and actuators. It has to deal with real physics, uncertainty and changing environments, and for this it needs real reference systems: digital twins, 3D capture, sensor data and real test environments.
Why real data counts
Synthetic simulations are predictable, reality is not. Real geometries and operating conditions provide the complexity needed for robust behaviour.
Limits
Physical AI is data-intensive and strongly dependent on the application context. Universally valid solutions only emerge step by step.
Sim-to-real describes the transfer of skills a robot has learned in simulation into real-world operation. The difference between simulation and reality is called the sim-to-real gap: in simulation many things work that fail in reality due to friction, sensor noise, tolerances and unexpected situations.
How the gap gets smaller
Through real 3D data, measured geometries, genuine operating conditions and tests in real training environments.
Limits
The gap can be reduced, but never fully closed. Real-world testing and safeguarding remain indispensable.
Why humanoid robots cannot simply be assessed like machines
Classic machines work within a defined, safeguarded framework. Humanoid robots move freely among people and take on changing tasks. What matters, therefore, is not capability alone, but whether their behaviour is sufficiently safe, traceable and controllable in a specific environment.
The central question
Not "What can the robot do?", but "Can it perform this task in this environment with these people safely and traceably?"
Consequence
An impressive demonstration is no substitute for an assessment of task, environment, risks, protective measures and responsibilities.
Difference between manufacturer and operator duties
The manufacturer places a safe, conforming machine on the market and provides technical documentation, a declaration of conformity and CE marking. The operator is responsible for safe use in the specific setting, including workplace risk assessment, instruction, servicing and maintenance.
Important to know
Anyone who substantially modifies a purchased, conforming system or interlinks several machines into one installation can themselves become the manufacturer in the sense of the regulations.
Limits
The exact delineation depends on the individual case and should, where necessary, be coordinated with specialist bodies.
Robotics projects rarely succeed on the drawing board, but through a clear, controllable start. These points help to assess a project realistically before investing.
The most important questions
Which concrete task should the robot take on, and in what quantity?
What does the environment look like, and which interfaces to existing production exist?
Which safety requirements and operator duties apply in the specific case?
Does the deployment pay off, and from what point in time?
How can the system be integrated into existing processes and data?
What documentation and what evidence are needed?
Is there a realistic step plan from pilot to productive operation?
