Through cognitive teaming to human-centred Industry 4.0

The Fraunhofer Institute for Machine Tools and Forming Technology IWU wants to answer the following questions: How do people work together with machines? How can digital assistants support employees in the factory without overburdening them with their complexity? How can technology based on the needs and strengths of people help to turn their own creativity into added value or help to secure the valuable experience of older employees and get them excited about innovative production technology? Dr habil. Franziska Bocklisch and her new group „Cognitive Teaming of Humans and Cyberphysical Production Systems“ are setting out to find the right answers.

Cognitive psychology analyses to advance production technology

With Franziska Bocklisch, the IWU is strengthening itself in a specialist area that at first glance seems unusual for an institute specialising in production technology - cognitive psychology. This deals with aspects of human thought and behaviour, such as perception, attention, memory and decision-making. The discipline investigates how sensory information is processed and turned into knowledge units. But also how this expert knowledge influences the interpretation of information and subsequent decisions. Cognitive psychology also explores the question of how humans can solve complex problems using creative strategies and reduce complexity in a meaningful way. A supposed human weakness when faced with an imminent challenge is actually a great strength - we reflect on the situation, place it in a broader context and draw on our experience. He asks himself what once helped him in a comparable, earlier situation to successfully master a task that at first glance seems almost impossible. An important area of application of the knowledge gained in this way is now aimed at modern production technology. One focus of the new working group at IWU is how the interaction between humans and technology can be made even more valuable in the future.

In future, people and technical systems will coach each other

Many innovations in the fields of robotics, artificial intelligence (AI), data analytics and visualisation technology are shaping modern industrial production (Industry 4.0), the report continues. AI is precise and repeatable. And thanks to its impressive computing power, it can now handle huge amounts of data in a meaningful way. At the same time, powerful assistance systems have sometimes even increased the complexity of human labour in production. The claim that technology can optimally support people and give them more freedom for more valuable creativity has therefore not yet been fully realised, the researchers note.

The new team at Fraunhofer IWU continues to focus on combining the respective strengths of humans and technology (complementarity), but emphasises the team concept because it is the prerequisite for a further quality boost in collaboration.

Two essential characteristics of teamwork are shared knowledge in order to achieve common goals. Easily understandable AI algorithms that match the structure of human expertise and the approach of experts in a particular field could then become real „cyber-helpers“. In a kind of reciprocal coaching, the technology then tells the human how a task can be solved even better. This can work, for example, by accessing well-structured, relevant data provided by an assistance system. Conversely, employees could, for example, make an AI solution that does not yet know all the decision options more stable. They can do this by understanding how it works and creating suitable decision bases, which the AI learns in a new training loop based on knowledge and data.

The technology should be thought of from the human perspective

If the claim to think about assistance systems from a human perspective is taken seriously, the key question should not be what is technically possible, but what can humans process well and what possibilities offer them real assistance. To return to the example of AI, it appears that with large numbers of units, the set-up and customisation costs of AI systems are proportionate to the support provided to employees. However, if the products change frequently, the maintenance effort is logically very high compared to the high number of units. Sometimes too high! For example, if a fallback solution is also required for the system failure. Experienced decision-makers and users therefore know that even the best AI is not error-free. But if you understand how it works, you can use it all the more profitably, the IWU researchers emphasise.

Systems that are too complex inhibit employee training

It is said that systems that are too complex make it difficult for new employees to familiarise themselves with them. However, if developed and used sensibly, „intelligent“ systems support the safeguarding of competences and the transfer of experiential knowledge. Initial research results on the observation of production processes by experienced employees, for example, are promising, the researchers note. The evaluation enables targeted enquiry and thus process descriptions of a much higher quality. This is especially true when it comes to communicating to new employees what is most important in their job.

The IWU has observed roll forming by robot, a multi-stage (incremental) sheet metal forming process suitable for small quantities. Ultimately, this type of roll forming was systematically described at various levels. An eye tracker followed the eyes of the technical experts and transferred them to a tablet, as the image shows. Recording and detailed analysis of the human gaze data then made it possible to ask specific questions: for what reasons was a particular process section carried out and why did certain aspects receive (special) attention? According to the scientists, this approach provides important starting points for developing cognitive assistance systems and goal-oriented automation systems.

Human decisions are supplemented by AI

In another research project at Chemnitz University of Technology, the research group „Human Cyber Physical Systems“ investigated how humans and AI can achieve common goals and acquire shared knowledge in a thermal coating process. The aim is to achieve high-quality surfaces while minimising material consumption. The team modelled the numerous setting options of a complex system for this purpose and the diverse quality objectives, which sometimes contradict each other, in harmony with human expertise. The human decision-making process is taken into account, precisely supplemented and not replaced. Because AI modelling can make complex patterns in the high-resolution technical process data easily understandable for humans, the first significant step towards team building between humans and the system has already been taken.

Adapting cyber-physical systems to the cognitive abilities of humans

It is important to anchor the teaming concept for better cooperation between humans and technology on a broad basis. Cognitive psychology expertise is rarely available in operational practice - and that doesn't have to be the case if science and industry work together across disciplines and institutions and, for example, students are already sensitised to designing technology more from a human perspective. The Fraunhofer IWU and the Faculty of Mechanical Engineering at Chemnitz University of Technology have been working together successfully on numerous research projects for many years. In the spirit of intensified cooperation, Franziska Bocklisch continues to head the „Human Cyber Physical Systems“ group at the Chair of Materials and Surface Engineering, where she contributes to the “Human-Machine-Teaming“ profile line of the Faculty of Mechanical Engineering.

Promoting a „real“ teaming of humans and production systems in basic and application-orientated research and preparing students for this field of work in the best possible way is a common incentive for both research institutions. Bocklisch concludes: „In technical development and also in operational implementation, the focus should be on people with their expertise and their needs. Our approach is to adapt cyber-physical systems to the cognitive abilities of humans - and not the other way round“.

Source: Press release Fraunhofer IWU In: blechnet from 19 May 2023