Trust layer required
Despite the euphoria, the path to widespread distribution is not a sure-fire success. The Bitkom paper makes it clear that there is still a need for improvement across all industries in order to make Physical AI a success on a broad scale. For example, a trust layer is needed. Concepts like the European Business Wallet are intended to ensure that every decision made by an AI is understandable and every identity of a robot is verified.
This is the only way to minimize safety risks and ensure acceptance in highly regulated industries such as the railways or the chemical industry. Additionally, Physical AI is more than just another application – it is a systems approach that inextricably weaves hardware and software together. If these systems can be integrated into the value chains in a secure and scalable manner, AI could actively help build the world of tomorrow.
More practical examples
The paper also gives further practical examples:
- SmartFactoryKL & XITASO (Agentic Production):
Products and machines organize themselves independently via autonomous software agents. The system adapts production planning to disruptions in real time to shorten throughput times.
- MaibornWolff (Assistenzrobotik):
An AI-supported robot supports employees in assembly tasks. With annual rental costs of around 93,000 euros, this saves around 69,000 euros per year and workplace compared to new staff.
- Airbus & XITASO (Project ENGEL):
An assistance system for rescue helicopters uses physical AI to detect obstacles during flight and calculate safe landing routes. The aim is to reduce operational flight time by around 10 percent.
