The convergence of artificial intelligence and traditional machine control as the basis for ‘Physical AI’ is a key topic in industry. For this reason, automation specialist Beckhoff will be demonstrating at the 2026 Hannover Messe how large language models (LLMs) can directly influence real-world motion sequences via standardised interfaces.
The integration of physical AI marks a paradigm shift in the manufacturing industry. Establishing a direct link between AI models and deterministic control technology allows machines to do more than just process static commands – it helps them create context-specific, autonomous responses to sophisticated requirements.
In February, Beckhoff provided an initial concrete insight into this technology at an event using a compact demo cell. The XPlanar planar motor system was combined with the TwinCAT CoAgent AI tool and an audio interface for voice commands. This illustrated a physical AI application in which levitating moving elements can be controlled via natural language and trigger the next movement sequence on command. This demonstration showed how simple and intuitive human-machine collaboration will be in the future, enabling even users without programming knowledge to carry out complex automation tasks.
Beckhoff will be scaling this scenario to a fully integrated industrial application at Hannover Messe 2026, taking place from April 20 to 24. The Atro modular industrial robot system, which is programmed and controlled using TwinCAT CoAgent for Operations using voice commands, will take center stage here. Based on the Model Context Protocol (MCP), the control system acts as an intelligent agent that translates human speech into machine commands, orchestrates path planning, and performs diagnostic tasks. The physical AI application will be showcased using a fun approach: the exhibit will play chess against visitors.
With tools such as TwinCAT CoAgent and TwinCAT Machine Learning Creator, Beckhoff already offers an ecosystem that facilitates this new era of automation. The tools support machine builders throughout the entire life cycle – from code generation in engineering through to error analysis during operation.
“We are moving AI away from chat windows and directly into machines and enabling language models to access the real world of controls through new standards such as MCP,” says Hans Beckhoff, Managing Director and owner of Beckhoff, describing this new development. “Technical inventions have always influenced the way that society as a whole evolves. Artificial intelligence and physical AI have significant implications, on the same level as the steam engine and electricity.”
Picture caption: Physical AI in practice: TwinCAT CoAgent translates natural language into machine commands and enables intuitive control of complex mechatronic systems.
Source: Beckhoff
