Robotics and AI Our robots are this intelligent

Related Vendors

Fraunhofer-Institut für Produktionstechnik und Automatisierung IPA

FAULHABER Drive Systems

Taiwan Machine Tool & Accessory Builders' Association (TMBA)

Wuxi InfiMotion Technology Co., Ltd.

In the industry, there is hardly any application that comes to market without an AI function. Robotics, in particular, could benefit from this boom. But what is already possible today?

ChatGPT, Perplexity, Claude, Deepseek: Artificial intelligence now has many names in the public domain and has arrived. In a very short time, AI assistants have conquered the world. In everyday life, users can turn to ChatGPT via an app and converse with the AI on any conceivable topic as if it were a human conversation partner. The breakthrough of generative AI has also caused a surge of innovation in the industry. Machines that can learn independently and make autonomous decisions are a game changer for the industry and robotics.

The rapid developments are also being monitored by Marco Huber, Scientific Director for Digitization and Artificial Intelligence at the Fraunhofer Institute for Manufacturing Engineering and Automation IPA. He is aware of the advantages that AI now offers to the industry: "Consider, for example, huge inventories from which robots are supposed to pick something. It is absolutely uneconomical to teach all objects manually. With the help of AI, this is no longer necessary." The key here is intelligent image processing. With its help, the robot recognizes completely unknown objects and can grasp them. This was not possible before. Tasks that only a human could perform are now suddenly feasible for a machine equipped with AI—and without time-consuming and uneconomical training efforts.

Numerous challenges

But there are also challenges that need to be addressed. "This starts with the appropriate digital infrastructure that must be in place and ends with the security and regulatory requirements. Depending on which AI method is used, the corresponding training data is required, and, of course, there must be a meaningful use case where AI can be deployed in a value-adding way," says Huber. Small and medium-sized enterprises, in particular, find it difficult to grasp the complex field of AI-related requirements. Legal questions also arise, evolving and changing constantly—keyword AI Act. The explainability of AI and the acceptance of employees also play an important role, according to Huber. Funding and information formats, such as the AI Progress Center initiated by Fraunhofer IPA and IAO and other training institutions, are intended to support companies.

And what's next? "Generative AI will further advance robotics, particularly through improvements in robot programming and path planning, as well as in optimizing energy efficiency and safety," says Huber. In production, AI could optimize machine parameters and bring production processes to an optimal state more quickly. What is already possible was demonstrated by Fraunhofer IPA with a demonstrator at Logimat in Stuttgart: With the help of AI, packing in a logistics process can be accelerated. According to the researchers, the demonstrator achieves 1300 picks per hour without prior knowledge of the objects. "Other examples are projects like our spin-off Cellios, which enables robot-based cable assembly. With it, robots can insert cables into connectors. The 16 to 18 openings of the connectors are as small as a fingernail—without AI, such a robotic application would be unthinkable," says Huber.

The next step in research is quantum machine learning: AI meets quantum computing. "However, further progress is primarily needed in quantum computing hardware to achieve fault-tolerant and scalable systems. Recent breakthroughs by Google or Microsoft, for example, are encouraging," explains the scientist.

Machine Authentication

Machine Operator Authentication - Why knowing "who" matters