For years, the Dutch industrial sector has struggled with rising labour shortages. During the next few years, large numbers of engineers and maintenance mechanics will reach retirement age. At the same time, not enough younger people with technical training are joining the labour market. Digital and Information Technologies, for example generative AI, large language models (LLMs) and digital twins, now offer the solution: a virtual service mechanic. This is a system that combines real-time data with error analyses and documentation to help mechanics diagnose and resolve problems remotely.
It is being researched by the Industrial Digital Twins research group of Amsterdam University of Applied Sciences (AUAS). “This will help us to solve a real societal and industrial problem.”
Challenge
Productivity rates in Dutch industrial sector depressed by labour shortages, virtual service mechanic offers solution
For years, the Dutch industrial sector has struggled with rising labour shortages. This problem is becoming more and more pressing, in both technology and manufacturing. According to recent data, more than 70% of businesses struggle to find enough qualified workers. These shortages have a direct impact on productivity. For example, the industrial sector’s revenue fell again last year, owing partly to the shortage of trained workers. This means higher workloads for existing employees, longer waiting times for customers and a decline in overall productivity.
All of these issues can be traced back in part to the ageing workforce. During the next few years, large numbers of engineers and maintenance mechanics will reach retirement age. At the same time, not enough younger people with technical training are joining the labour market. The problem is made worse by the declining popularity of technical education among young people. As jobs in technology become increasingly specialised, young people are more attracted to other sectors such as the creative and service industries. The result is a structural supply and demand mismatch on the labour market – a mismatch that will only become worse during the years ahead.
The consequences of these shortages are reflected in declining labour productivity rates for the industrial sector, with last year showing the greatest drop for nearly fifteen years. It is becoming clear that the Dutch industrial sector urgently needs new solutions to overcome these problems. This is where generative AI, large language models (LLMs) and digital twins offer a solution.
AUAS research professor in Industrial Digital Twins Jurjen Helmus explains,
“Generative AI is being used for plenty of pointless purposes – such as to generate images or videos of all kinds of things. The virtual service mechanic, however, puts generative AI towards solving a real societal and industrial problem.”
Results
Virtual service mechanic: a groundbreaking solution
To relieve the growing pressure on the industrial sector, AUAS’s Industrial Digital Twins research group and the companies IJskoud and Freezerdata have developed an innovative solution: the virtual service mechanic. This is a system that combines real-time data from machines with error analyses and access to the machines’ service manuals, for a remote mechanic to quickly and effectively diagnose and resolve the problem.
The system integrates advanced language models, underlying service information and real-time analyses to allow an inexperienced mechanic to talk to the machine. For example, they could simply ask, “How are you doing?” The machine can respond by announcing straight away whether anything is wrong, or what is functioning properly. The system is capable of retrieving underlying information from manuals and work tickets, regardless of language, while also communicating in the mechanic’s own language. This makes it the perfect AI buddy for mechanics, assisting them with their jobs.
What makes the virtual service mechanic so special is the possibility to adapt to the user’s experience level: less experienced mechanics receive more help, while experienced engineers can manage with less assistance. The project also includes using a language model for what it does best: generating language.
By separating language and information, the virtual service mechanic can communicate with the end user in any language they want. As a result, the system not only provides a solution to labour shortages, but also helps to make optimum use of the knowledge and skills of mechanics in other countries.
Click here to download the full project description (in Dutch only)