IFA 2026 in Berlin placed robots in front of a broad public rather than keeping them inside laboratories or specialist industrial events. From September 4 to 8, the show presented humanoids, quadrupeds, autonomous robot football and concepts for domestic assistance. Together, the demonstrations offered a visible snapshot of Physical AI: artificial intelligence connected to machines that perceive and act in the physical world.
The displays were ambitious, but they need careful interpretation. A robot performing on a runway or competing on a controlled pitch shows progress in movement, perception and interaction. It does not by itself prove commercial readiness, reliable operation in a home or mass adoption. IFA’s value was in making both the possibilities and the remaining gap easier to see.
Physical AI moves onto the public stage
Physical AI extends machine intelligence beyond text, images and software interfaces. A physical system must understand its surroundings, respond to change and carry out actions through hardware. That adds constraints that a purely digital model does not face: balance, friction, battery life, sensor noise, unpredictable objects and the need to operate safely around people.
NEURA Robotics used its IFA keynote to describe Physical AI as a bridge between AI systems and machines capable of perceiving and acting. The company connected the idea to industrial needs, demographic change and labour shortages. Those are important motivations, although potential social or economic value does not remove the engineering work needed for dependable deployment.
IFA is particularly relevant because it reaches consumers as well as specialists. Visitors could see robots in contexts associated with entertainment, emergencies, kitchens and homes. That public exposure helps move the discussion from abstract AI capability to questions about what a machine can physically do, how people interact with it and where trust is required.
Humanoids demonstrate agility and interaction
The “Robots on the Runway” programme featured dancing, Tai Chi, backflips and other movement demonstrations. Robots also answered questions and interacted with the audience. IFA listed participants including PrimeBot Robots, EngineAI, Cozio Robotics, Dobot Robot, DEEP Robotics, Agibot, Unitree, ZBy, Ollobot, Astrall Dynamics and AIMOGA.
These performances reveal advances in balance, control and coordinated motion. They are also designed for a predictable stage. Industrial and domestic tasks introduce clutter, unexpected contact, changing surfaces and long operating periods. A successful demonstration is therefore evidence of a capability under defined conditions, not proof that the same system can perform useful work all day without supervision.
The range of exhibitors nevertheless matters. Humanoid development is no longer limited to one or two highly visible laboratories. Multiple companies are pursuing different combinations of hardware, control software and intended markets. That diversity can accelerate learning, but it also makes direct comparisons difficult because demonstrations and performance claims do not follow one common standard.
Robots for emergencies, homes and kitchens
IFA also showed a four-legged robot presented as a possible tool for fires and other emergencies. Quadrupeds can move across terrain that is difficult for wheeled machines and carry cameras or sensors into dangerous areas. The practical question is how reliably they operate under heat, smoke, debris, poor connectivity and other conditions far more demanding than an exhibition floor.
In a panel on robots in kitchens and homes, experts discussed a shift from task-specific appliances toward systems that could understand a household’s environment, habits and preferences. A small humanoid was brought onto the stage as an illustration of that direction. The concept points toward more adaptable assistance, but private homes are among the hardest environments for robotics because layouts, objects and human expectations vary widely.
Domestic robots will also face requirements beyond technical capability. Privacy, local data processing, affordability, maintenance and simple recovery from errors will shape adoption. A machine that works only when a specialist is nearby is very different from a product that families can safely use every day.
Autonomous systems take the football pitch
RoboCup events at IFA brought autonomous machines onto a football pitch from September 4 to 8. Teams of researchers and students developed systems that analyse the game, make decisions and coordinate with teammates. Robot football concentrates several difficult problems into a clear environment: perception, localization, planning, communication and movement under changing conditions.
Competition can be valuable because goals and failures are visible. A robot must respond in real time rather than replay a fixed routine. Even so, a sports field is structured and the rules are known. Skills developed there can contribute to broader robotics research, but transferring them to warehouses, streets or homes requires further engineering and validation.
The educational role is also important. Competitions give students a way to combine computer science, electrical engineering, mechanical engineering and AI. They create shared benchmarks and expose weaknesses that may be hidden in isolated laboratory tests.
Demonstration is not deployment
IFA 2026 showed that robots are becoming more capable and more visible. It did not establish that every displayed concept is ready for sale or independent everyday operation. Deployment requires durability, predictable safety, service networks, software support and an economic case. These factors are less dramatic than a backflip but more important to users.
Public demonstrations can also compress complex technology into a short performance. Viewers may not see remote assistance, preparation time, environmental controls or the number of unsuccessful attempts. Responsible reporting should therefore separate what happened on stage from broader claims about autonomy and readiness.
The most informative next steps will be deployments with measurable outcomes: hours of operation, intervention rates, task success, maintenance needs and performance in changing conditions. Those data will show whether Physical AI is moving from impressive prototypes toward dependable tools.
RoviVox View
IFA 2026 made Physical AI tangible. Humanoids, quadrupeds and autonomous competitors showed that perception and movement are advancing across a wider field of companies and research teams. The strongest signal is not any single stunt, but the growing number of systems being tested in public.
At the same time, the exhibition underlined the difference between capability and product maturity. Everyday robots will succeed when they are safe, reliable, maintainable and useful without constant expert support. IFA showed the direction of travel; real homes, workplaces and emergency environments will determine the pace.