Adlershof Innovation Booster

Thanks to AI, new technologies are emerging on campus that are driving scientific excellence and strengthening technological autonomy

Artificial intelligence is accelerating research and driving innovation cycles. In Adlershof, this is giving rise to new technologies that not only advance scientific excellence but also strengthen the technological autonomy of Germany and Europe.

The Helmholtz Centre Berlin for Materials and Energy (HZB) is demonstrating just how transformative artificial intelligence can be. With the new department AI and Biomolecular Structures, the institute is bringing AI into structural biology. Biophysicist Andrea Thorn has been building up the department within the Photon Science division since July 2025. “The aim is to evaluate large experimental datasets more effectively and significantly accelerate progress in the life sciences,” Thorn says, who has been working with machine learning as a researcher for more than a decade. “We are developing methods and software that combine complex biomolecular data from different sources in meaningful ways,” she says. The work builds on measurements taken at major research facilities such as the MX beamlines at BESSY II, complemented by datasets from synchrotron sources, electron microscopy and neutron facilities worldwide. “More than 200,000 biomolecular structures have now been published,” Thorn notes. “That is simply too much information to combine manually.”

Her department was established to make this flood of data better usable. “On average, there is a discrepancy of around 24 percent between our established structural models and the measured data,” says Thorn. AI-based and statistical approaches could help improve both data quality and the accuracy of structural models, while also making it easier to systematically integrate results from various methods.

In the long term, the team is pursuing an ambitious goal: “A system like this could significantly advance computer-aided drug design—a type of ChatGPT for molecules—that brings together structural-biological information from experiments, databases and scientific literature and presents it in an accessible way.” For now, that vision remains a long-term objective. The immediate priority is to establish machine learning as an intelligent support system for researchers.

A particular focus is on explainable AI. “AI-generated results must be transparent and scientifically verifiable,” says Thorn. New scientific insights can only emerge when it is clear exactly why a model has arrived at a particular conclusion.

The department’s work also contributes to the technological sovereignty of Germany and Europe. Independently developed methods, open scientific data and close integration within the Helmholtz research ecosystem help ensure independence from external platforms. In this way, Adlershof is strengthening its role as a location where AI is developed responsibly and applied productively in research.

Developing reliable AI for industrial use will also be crucial for technological sovereignty in the years ahead. The use of artificial intelligence has long been seen as a key driver of efficiency in industrial quality assurance and non-destructive testing (NDT). In practice, however, many projects fail. “Most industrial AI projects don’t fail during model training,” says Kemal Levi, founder and CEO of Relimetrics. “They fail because the systems prove too fragile in real-world production environments, too difficult to adapt, and too expensive to operate.”

This is precisely where the Adlershof-based company comes in. Its platform ReliVision addresses one of the central challenges of industrial AI: reliability in continuous operation and under changing conditions—such as new products, suppliers, defect types or regulatory requirements. “ReliVision allows companies to build, validate and operate AI-based inspection systems on their own infrastructure,” Levi explains. Domain expertise from the users themselves is integrated into the system. The result is faster time-to-value, reduced dependence on external integrators, and AI that becomes a lasting operational capability rather than a one-off pilot project.

Hardware independence is another decisive factor. Industrial production environments often evolve over many years and are technologically heterogeneous. “Hardware dependence makes AI systems inflexible and expensive to scale,” Levi says. As a software-defined solution, ReliVision allows existing cameras, sensors and computing systems to continue to be used—even across different manufacturers. Levi: “This protects investments and makes it easier to roll out AI across multiple production lines and sites.”

Technological sovereignty is equally central to the concept. ReliVision runs entirely on-premise, meaning that data, models and decision logic remain with the customer. Sovereignty is not just about where data is stored,” Levi says with emphasis. “It is also about who controls the training, validation, auditing and further development of the AI.” Especially when dealing with sensitive production data and proprietary know-how, this becomes a decisive competitive advantage.

Levi believes Adlershof offers ideal conditions for such developments. The close interaction between applied research, industrial relevance and engineering expertise creates an environment in which robust AI systems for real-world use can emerge. This ecosystem has also enabled Relimetrics to work closely with application-oriented research partners such as the Fraunhofer Institute for Ceramic Technologies and Systems (IKTS), translating advanced research in inspection and materials science into deployable industrial AI solutions.

Artificial intelligence is also set to transform agriculture. Removing weeds with targeted laser pulses—without chemicals or labour-intensive manual work? Companies such as Laser Electronics LE GmbH in Adlershof are already making it possible. The company demonstrates how industrial photonics can enable AI-driven applications in smart farming—and how technological sovereignty “made in Germany” can be created.

Founded in 2003, the company is one of the leading manufacturers of laser-diode controllers, drivers and complete laser systems. Its products are used worldwide in industry, medical technology as well as research and development. “We have been based at the technology campus for ten years,” says founder and managing director Ralf Ziegler. “During that time we have steadily expanded our expertise, particularly in the field of blue diode lasers.”

One visible outcome of this work is the BluEX Laser Weeding Tool. Using precise blue laser pulses in the spectral range of 450 nanometres, weeds can be destroyed with high accuracy. “This conserves resources, reduces the need for herbicides, and opens up new possibilities for sustainable agriculture,” Ziegler explains. The modular system is designed to be mounted on tractors or autonomous platforms and operates together with cameras and AI software from other manufacturers. Image data from the field is analysed using AI to distinguish crops from weeds, after which the laser is directed with pinpoint precision. Laser Electronics collaborates with specialised partners for AI, sensor technology and vehicle platforms, providing the core of the system—the robust, energy-efficient high-power laser together with cooling, safety and beam-steering technology.

A prototype is scheduled for quite literal field trials in spring 2026, after just two years of development. “Short distances, specialised neighbouring companies in mechanics, optics and electronics, and strong international connections allow for very fast development cycles,” says development manager Michael Häußler. This is precisely the kind of ecosystem that strengthens technological sovereignty. Key technologies such as efficient blue lasers are developed and produced locally and transferred directly into new applications. In doing so, Adlershof becomes a catalyst for sustainable innovation—and Laser Electronics an example of how high-tech engineering from Germany can help shape global markets.

Chris Löwer for Adlershof Journal

• New department at HZB: ‘AI and Biomolecular Structures’ - Helmholtz-Zentrum Berlin (HZB)
• ReliVision: Unlock the Power of AI for Industrial Automation | Relimetrics
• Laser Electronics