AI in Production

Interview with Sven Hamann, Managing Director at Bosch Connected Industry

Artificial intelligence is considered a key technology for industrial applications. Sven Hamann, Managing Director of Bosch Connected Industry explains the concrete benefits for manufacturing companies.

© Bosch
Industrie 4.0 is coming of age – thanks in part to Bosch's pioneering work in AI.

futur:  A Germany-wide survey by Bosch indicates that AI is highly popular as a key technology for industrial applications. What hopes do people associate with it?


We do indeed sense that the industrial application of AI is currently gaining momentum. There is a great deal of curiosity and openness towards trying out AI. Many are hoping for greater efficiency, better work results, or greater safety in the operation of plants. Others see the technology as a driver of complexity and are concerned about a loss of control. That is why I am convinced that AI has to prove how it can be beneficial. We see this in products in the consumer segment: In this area, AI has now reached a level of maturity that provides a high level of benefit, and where I, as a user, tend to experience a reduction in complexity, such as in speech recognition.

I believe that AI will also gain a lot of momentum in industrial production. In our plants, where we integrated AI solutions into specific use cases at a very early stage, we now see that we can leverage enormous potential. We are talking about savings amounting to several million euros per plant. These are pivotal aspects that will profoundly accelerate the introduction of AI in companies.

futur: Which specific AI applications are already being used at Bosch?


In manufacturing, we are already employing AI in a number of places. The classic example is certainly predictive maintenance. In this scenario, the current condition of the machine is monitored via sensor data. We use the data to detect malfunctions even before a production standstill can occur. By analyzing data from our manufacturing processes, we can gain new knowledge regarding the production processes. We then use this knowledge to optimize production parameters or cycle times. We also use AI in intralogistics. In a pilot project, we are currently optimizing the material supply dynamically during ongoing operation, thereby constantly adap-ting it to current conditions.

futur: Despite its many advantages companies often still find it difficult to use AI technology. In your view, what are the biggest obstacles?


One obstacle is in fact the degree of digitalization itself, which still varies greatly from one industry and company to another. The very first step is the availability of data throughout the entire life cycle of a product, i.e., starting with the product creation process, design, development, production, and ultimately its operation. This data must be prepared in such a way that makes it meaningful. If it is enriched with semantic data structures, for example, companies can tap its potential in two directions. For one, the feedback loop can be used to optimize the next generation of products, or to adapt products to specific user behavior. The second is to tap into business models. This constitutes the foundation. Before that, it is quite difficult to use methods like AI. This groundwork is at different stages of completion in various companies.

Another obstacle is the availability of expertise. Artificial intelligence, as previously mentioned, carries a very high potential. However, its benefits only emerge in the domain. This means that companies need to bring together domain experts such as materials scientists or production engineers with AI experts. I think it is key to actually view this digital transformation as being a transformational project. It starts with creating transparency – concerning the goals I am pursuing, about planning and implementation – and giving employees the opportunity to get involved, help shape things, and further develop themselves.  The key aspect here being lifelong learning.

»Our goal is that by 2025, all Bosch products will contain AI or at least be developed and produced using AI.«

– Sven Hamann

futur:  How does Bosch succeed in combining its own domain know-how with AI expertise?


We try to do this deliberately in our Bosch Center for Artificial Intelligence. Here, we bundle the AI expertise of almost 300 people from different locations in seven countries. This allows us to involve AI experts at a very early stage of product or process development. The advantage for us, of course, is that the robots and machines are in the labs just next door. This means we have a very rapid feedback loop and can quickly identify what benefits an application will bring. Naturally, this is highly attractive for someone who wants to see the effects they are achieving with their work.

futur: How important are partnerships in building such expertise? 


Extremely. The idea that you can do things entirely on your own has become outdated and is no longer realistic, especially in light of the complexity of the topics. We therefore make it a point to establish partnerships.

One such example is Cyber Valley in Tübingen, where we are conducting joint research with partners from the academic field and the private sector in the areas of AI, machine learning, robotics, and computer vision. On the one hand, we are faced with the challenge of applying the technologies to concrete use cases and generating benefits. But on the other hand, there are fundamental questions that have not yet been solved. One example is »Understandable AI«: How can I verify AI-based systems? How can I reduce the number of training cycles for my algorithms and still achieve a high-quality result? This is relevant in, e.g., potentially critical use cases such as autonomous driving.

In production, it is all about quality, which is of course also a valuable commodity for us suppliers. Manufacturers require an intelligent system the most when they introduce a new product. In order to start using the appropriate production processes, they need quality assurance. However, they do not yet have any training data at all to do so. This is a basic problem that we are very much concerned with. Furthermore, the robustness and transferability of the AI models to different machines and plants is also another issue. These are all fundamental challenges that cannot be solved by a company working on its own. That is why collaborative research work is necessary.

futur: How long do you think it will take for AI to become a standard tool in companies?


That is difficult to answer in such general terms. But I can tell you very specifically what we have set out to do: Our goal is that by 2025, all Bosch products, across the entire range and spectrum, will contain AI or at least be developed and produced using AI. We are implementing this step by step, and are well on our way.

Sven Hamann

© Bosch

Managing Director at Bosch Connected Industry, Robert  Bosch Manufacturing Solutions GmbH

Sven Hamann has headed the Bosch Connected Industry business unit since July 1, 2019. Before that, he was responsible for the central research area for manufacturing automation and metal and plastics technology at  Bosch. He has broad international experience in production and mechanical engineering and holds a Diploma degree in engineering from the Technical University of Berlin with a focus on information technology in mechanical engineering.