October 6, 2026 | Machine learning, robotics, logistics and Industry 4.0

Quantum AI for classical robots

In future, artificial intelligence on devices with conventional hardware should be able to benefit from the highly efficient training of quantum AI
DLR has developed a method that transfers a trained quantum AI from a quantum computer to the classical hardware of a robot. As a first practical example, the project team used it to train a model robot.
  • DLR has developed a method that makes it possible to use the training advantages of quantum AI on classical hardware.
  • To do this, DLR researchers modified a quantum-trained AI so that it can run on conventional computers and then transferred it to a robot.
  • In future, highly efficient quantum algorithms could simplify and accelerate AI training, with only the training result being transferred to end devices.
  • Focus: Artificial intelligence (AI), quantum AI, quantum computing, machine learning, intelligent robotics, transport, logistics, Industry 4.0, security, cybersecurity

The German Aerospace Center (Deutsches Zentrum für Luft- und Raumfahrt; DLR) has developed a method that makes it possible to use the training benefits of future quantum artificial intelligence (quantum AI) on classical, embedded hardware too, for example in robotics. Quantum AI can learn faster and more effectively than classical AI. DLR researchers have succeeded in transferring the training result of a quantum AI to the hardware of a conventional robot. This opens up new, highly efficient and more cost-effective approaches for using AI in complex and dynamic tasks, such as machine learning, logistics and Industry 4.0.

At Quantum Effects 2026 (Messe Stuttgart, Hall 2, Stand 2E17), the DLR Institute for AI Safety and Security is demonstrating how a 'translated' quantum AI can be successfully deployed in a robot with a gripper arm.

Training – AI's store of knowledge

The better an AI is trained, the greater its store of knowledge and the more reliably it can make decisions. Training determines the AI's performance and, at the same time, poses the greatest challenge for its operational safety. Classical AI requires large amounts of data and many training runs. Quantum computers can accelerate and simplify this learning process, saving energy, time and data – and ultimately costs.

Quantum AI has great application potential. It can handle complex and dynamic tasks, even in constantly changing environments. Potential applications exist wherever a high degree of adaptability or interaction with people is required, including transporting materials, loading machines, gripping complex objects, using tools, carrying out assembly work and controlling vehicles. Humanoid robotics, in particular, is a future field of application, with uses ranging from care and medicine to planetary exploration missions.

Reaping the benefits of future quantum AI today

"Quantum AI only works on quantum computers. However, a universally deployable, error-free quantum computer does not exist yet. Even if these computers become available in the future, not every AI-controlled end device will be equipped with one", explains Lea Schmitz from the DLR Institute for AI Safety and Security.

To make the training advantage of future quantum AI usable on classical computers and on devices with simple hardware, the institute has developed and tested a multi-stage transfer method. This method translates the training result of a quantum AI into an optimised form that can be run on conventional hardware, making it possible to use AI on small, mobile devices with simple control computers.

"Our goal is to be able to apply quantum-trained AI independently of quantum computers. We see great potential, for example, in the automation of industrial processes or in robots – whose computing power is usually well below that of a standard PC – as well as in smart devices", says Schmitz.

Robot changes a wheel with 'translated' quantum AI

DLR researchers have now put the transfer concept into practice for the first time, showing that quantum AI can be transferred to simple, lightweight standard hardware. This is a crucial step towards making practical use of the possible advantages of quantum AI. "We are demonstrating this using a small mobile robot that grips a miniature tyre", explains their colleague Lautaro Hickmann. "We trained the quantum AI in advance using a simulation of the robot. To do this, we combined the robot movements from a classical simulation with control by the quantum AI."

The method is particularly suitable for simple, lightweight hardware in mobile robots, cranes or rovers. Another future field of application is small devices in the Internet of Things, ranging from smart, networked household devices equipped with small computers through to building technology. Here, the greatest challenge is keeping the required computing power and energy consumption low. In principle, the transfer method can be applied wherever AI is used or is set to be used.

Video: Training results from quantum AI for classical computers
Quantum computers can accelerate and simplify the AI learning process. The DLR Institute for AI Safety and Security has developed a method that allows the training results from a quantum AI to be transferred to devices equipped with classical computers. As a first practical example, the DLR project team used this method to train a model robot.
 
The applications in industry are diverse, such as for the automation of industrial processes or use in mobile robots, cranes, or rovers. These devices can operate autonomously without a data connection and are protected against cyberattacks.

Tensor networks translate quantum AI for classical hardware

The DLR project team first trained the quantum AI. However, it was not yet possible to carry out the training for their robot on currently available quantum computers, as the movements of the gripper arm, with its five joints, are too complex a task for them. The team therefore used a software simulator of a quantum computer. "When we simulate quantum physical principles such as superposition and entanglement on a classical computer, we naturally cannot achieve the high efficiency and speed of a real quantum computer. But we can already reproduce the fundamental principles", emphasises Hickmann.

Using what are known as tensor networks, all the quantum AI's training information is translated into a form that can be run on classical computers. This consists of mathematical structures that describe geometric and physical properties. To deploy the exported quantum AI on small devices as well, the network is also compressed and simplified. "Our method is already complete and can be used as soon as quantum hardware for training becomes available. That is when the process will become industrially relevant", adds Schmitz.

Quantum AI remains under lock and key

One security advantage of the transfer method is that the quantum AI can be trained centrally in a data centre or company, where the fully trained quantum AI remains securely locked away. Only the training result is specifically optimised for, and provided to, the end devices.

In addition, these end devices do not require a real-time connection to a quantum computer in the cloud. They can operate autonomously without a data connection and are therefore protected against cyberattacks or data theft. This is particularly relevant for series production. It also protects data sovereignty and enables quantum AI to be deployed on a large scale.

This research success marks an important step towards making quantum AI usable in practice. The next step is to work with industry to adapt the new method to specific use cases and industrial hardware, and to develop it further.

Related links

Contact

Jens Mende

Corporate Communications Stuttgart and Ulm
German Aerospace Center (DLR)
Corporate Communications
Pfaffenwaldring 38-40, 70569 Stuttgart
Tel: +49 0711 6862-229