Getting started with Artificial Intelligence
Artificial Intelligence (AI) offers opportunities for the (manufacturing) industry. But where to start with this technology? Entrepreneurs were inspired on Wednesday October 13 at Perron038, during the session ‘Getting started with Artificial Intelligence’. Experts from the field told about their experiments and successes in the field of AI. Perron038 presented a case from the project PRISMA.
Entrepreneurs in the East of the Netherlands can earn as much as 13 billion Euros in 10 years by making smart use of AI, according to figures from Buck Consultants International (BCI). But as cryptic as this sounds, in reality, it is just as complicated for many companies. Based on a clear social mission, Oost NL helps entrepreneurs in East Netherlands to develop their innovations to the next level. So that today’s technology can be used to create sustainable solutions for the future. This is why Oost NL, together with The Economic Board Arnhem Nijmegen, Twente Board, Radboud University, Radboudumc, University of Twente, Wageningen University & Research and Th!nk East Netherlands, founded the ‘AI Hub Oost-Nederland’.
AI Hub for Eastern Netherlands
Chris Willemsen, who is responsible for Business Development Digital Industries and the AI Hub Oost-Nederland, explains what the hub entails. In the East Netherlands, more than 350 companies are working on AI applications. In addition, three universities have a total of 1,100 scientists working on AI, and there are 5,700 university students and 8,700 college students each year. These separate links are brought together in the AI Hub Oost-Nederland. ’
Besides this hub, there are six other regional AI hubs. Together, they form the regional eyes and ears of the Dutch AI Coalition network. “We share stories about companies that are working on AI with the rest of the country. And vice versa,” Willemsen emphasises, “we pass on the European and national developments to the entrepreneurs here.
Furthermore, the regional hub is there to connect knowledge and issues. ‘In addition, we want to inspire, make talent visible and bring the subject of AI to the attention of management. What AI means and what they can do with it is not quite clear to them. Therefore, we focus not only on the technology, but also on the people.’
Parcel picking with robots
Wilbert van de Ridder, Development Engineer at AWL-Techniek and lecturer and researcher at Windesheim University of Applied Science, tells the audience about the application of Vision Technology in the PRISMA project. At Perron038 we have been working for almost two years on parcel picking by robots. The challenge in this case is that robots have to recognize different types of parcels, detect the right gripping point and then make the choice which parcel to pick up first. We decided to tackle this issue with vision technology. One of the lessons learned by Van de Ridder is the importance of source data. No matter how intelligent today’s technology is, a robot is not a human being. The robot can only perform what the human has programmed. “So, we must teach robots with each passing parcel how to handle it.”
One question that arises from this successful experiment is what AWL will do next with the developed technology. ‘Are we going to use it for our own work or are we going to look at commercial opportunities? That’s the question we’re working on now’, Van de Ridder concludes.
Rich inspection data
AWL had all types of packages available, but the case of Salland Engineering shows that it is not always that ‘easy’. There is often a lack of data to train AI algorithms. This can be solved with synthetic data, data that can be generated artificially. Salland Engineering is a market leader in the field of chip testing and tests for Bronkhorst, among others. Until now, this has always been done by people. But if robots can learn to recognise possible deviations in chips, they can take this repetitive work off people’s hands. Aleksandar Andreski, associate professor of Nanophysics at Saxion, explains how synthetic data and Machine Learning were used to detect defects in chips at an early stage. ‘The complexity lies in the fact that on the one hand there were rarely any faults in the chips, but on the other hand there were a great many different deviations possible. We designed all those defects ourselves because there were not many real examples. So we created a ‘large training set’ with synthetic data to feed the algorithms with. The result is ‘rich inspection data’. This enables us to know not just whether a chip is broken, but also what is wrong with it. That is useful information for the production process.’
Vasos Arnaoutis is the third speaker to highlight a practical case study. The researcher from the Frauenhofer Project Center at the University of Twente looked at the partial automation of the creative design process. In this case it concerns the design of moulds for product assembly, aimed at several production companies. The design of these moulds is a creative process. Can it be outsourced to software completely? As Arnaoutis puts it: ‘The ‘stupid’ part of design is. The people who would normally do that part of the design are now only sketching the framework. And that is the more complex work’, he emphasises.
This immediately answers the question that has been hanging in the air all afternoon: will AI replace people? No, people must feed the system, so that the system can help people. This way of thinking could well cause AI to come to imply: Assisted Intelligence.
The setting for the inspiring afternoon was Perron038, a place for the innovative manufacturing industry in the Zwolle region. An old industrial hall in the Zwolle station area has been converted into an ultramodern environment. This is where technological innovation comes about, as high-tech companies and educational and research institutions meet. Technicians and enthusiastic students work together and carry out research and experiments.
Companies working with, or getting started with AI are welcome to contact Chris Willemsen, AI Hub coordinator for East Netherlands: +31 6 2885 2115 or email@example.com. Or check out Aihub-oost.nl for more information.
Credits text: Mirjam van Huet