Automation "The copilot is a huge efficiency booster"

Related Vendors

Microsoft Deutschland GmbH

FAULHABER Drive Systems

Wuxi InfiMotion Technology Co., Ltd.

Together with industrial partners, Microsoft has implemented the copilot in both engineering and manufacturing. Nico Hartmann, Manufacturing Industry Lead at Microsoft, knows why the German industry offers the best prerequisites for this.

According to data from the International Monetary Fund (IMF), Germany is one of the three countries worldwide that can benefit the most from the use of artificial intelligence (AI). Why does Germany in particular benefit so much from the use of artificial intelligence?

Nico Hartmann: When it comes to the evaluation, the International Monetary Fund places a decisive role on a country's economic strength and infrastructural prerequisites. In basic research on artificial intelligence (AI), Germany has a very efficient basis with various AI clusters. The Technical University of Munich alone has one of the strongest AI clusters, including a potent start-up cluster. Others exist in Tübingen, Dresden, Heilbronn or Saarbrücken. At the same time, we have leading companies in Germany that also invest in AI and tailor AI applications to the manufacturing industry.

Microsoft provides a broad portfolio of tools for the use of AI in the industry, which partners like Siemens use to create industry-specific solutions. This is where the extensive domain knowledge from industrial manufacturing in Germany comes in and leads to state-of-the-art solutions that advance the industry. This is why we are perfectly positioned in Germany for the use of artificial intelligence.

How can Microsoft's copilots and AI models be integrated into one's own organization?

Firstly, we use our co-pilots in our modern work solutions, which are called Microsoft 365 today - so practically at every office workstation. In the industry, we offer the co-pilots as solutions of our partners directly, also via our marketplace. In addition, we have a modular system with over 1,600 different AI models. These are not exclusively Microsoft AI models, but also models from Open AI or models from other partners like Mistral or Data Bricks, who provide their basic models via us. We have created an open system. With these models, every company is able to develop its own AI applications for different business models.

Can you give a specific example of the integration of generative AI in the industry?

A strong example is the Siemens Industrial Copilot, which serves both development and manufacturing. In the background, our Azure Open AI service is used, so that users have the possibility to develop new CADs via natural language, for example, or to optimize existing ones, such as the space utilization of existing areas, assembly, manufacturability, or part availability. The corresponding PLM and CAD systems are running in the background in the cloud. Additional AI models interlock the applications with Siemens' proprietary systems, but also those of third parties.

In addition, the future will mainly be about advancing automation with the Industrial Copilot. In the past, automation mainly involved programming controls, such as the Siemens S7, or teaching robots, as well as simulation. Now, simulation is integrated and accelerated via High Performance Computing. Based on the CAD data, the entire production environment can be simulated and then the programming of the controller can be automated in the TIA Portal via the copilot. We always talk about copilots, not autopilots. The copilot makes suggestions, but the decision still lies with the engineer. However, the copilot takes over many administrative and documentation tasks.

How is AI advancing robotics?

Robotics, for example, benefits from the use of AI in teaching the robot. In the past, this was generally a process that dragged on for days because every single movement was defined. Now there are learning models based on the user or commissioning engineer's input to optimize paths or grip different part spectra. This accelerates and automates programming and commissioning. Adjustments can also be made much more flexibly, for example when new programs are implemented or processes are optimized. Because natural language input is used, the copilot is increasingly enabling people to make these improvements without having programming knowledge.

Humans make decisions, not artificial intelligence.

Nico Hartmann

You deliberately distinguish between the co-pilot and an autopilot – why?

Because the goal is not to enable AI to make fully automated decisions on its own. Instead, it should make suggestions, with the actual decision still being made by humans. The more critical the infrastructure, the more important it is that humans can intervene. At Microsoft, we place very high value on ethical standards for artificial intelligence and welcome the Artificial Intelligence Act of the European Union. Our own standards go far beyond the decisions at the European level, but the European AI law is important for creating trust. We want to see our technology used for good purposes and not put into the hands of someone who can cause harm with it.

At the Hannover Messe, Microsoft showed a showcase with BMW.  How does the co-pilot help there?

The use case at BMW showed the use of AI across the entire chain, starting from development to production. BMW uses AI to develop the systems on which the company will later manufacture the cars. In the design of these systems, a wide variety of AI use cases are relevant, from the configuration of the system to the documentation or derivation of training materials. Afterwards, this virtual environment is simulated. Here, high performance computing helps to get optimized results more quickly than in the past. Based on the simulation, the actual system is defined.

In this case, the scenario was transferred to Nvidia's Omniverse. Later, in operation, everything that happens in the plant can be rendered and visualized based solely on the operating data - and this in real time. The IoT data is consolidated and prequalified and contextualized via Microsoft's Manufacturing Data Solutions. This means that it is defined what kind of data they are, whether they are on the edge or in the cloud, and what data they represent.

Placed in the respective context, they are available for evaluation, which then becomes relevant for the metaverse in the next step. Here we usually talk about applications of the Hololens, but it can also be another device. The shopfloor worker can access all data with the Hololens. He stands at the plant and can ask the Microsoft Copilot Template for Factory Operations about the current status of the plant. This example shows: The broader the use of AI, the more the company benefits.

So, using the copilot also saves a lot of time?

One of our most fascinating applications was realized with Harting. This case shows how quickly we can move from idea to realization. At SPS in November 2023, the idea for a Generative Engineering Case was born, a configuration based on natural language input. In the background, Siemens Teamcenter and Siemens NX are running. Harting uses our AI to optimize CADs using natural language. In a very short time, about four months later, we were able to show the application live at Hannover Fair 2024.

What do you see as the greatest added value provided by the use of AI?

All solutions basically address the shortage of skilled workers, because artificial intelligence enables people to perform value-adding activities more effectively and efficiently. Therefore, we will definitely see an enormous efficiency boost in all areas of business, starting with engineering, through production and automation, all the way to after-sales.