Machine vision is an example of applied science, where the methods are often directly useful in commercially interesting applications – in contrast to physics or astronomy, for example. It therefore does not make sense to develop two separate branches of machine vision: academia and industry; instead, the two have to work together.
The European Machine Vision Association (EMVA) held its first European machine vision forum – from 8 to 9 September 2016 in Heidelberg, Germany – expressly to foster this collaboration between academia and industry.
Forming a link between the research and commercial worlds is mutually beneficial to both areas. Michael Heizmann, a professor at the Karlsruhe Institute of Technology (KIT), is also a consultant to the Fraunhofer Institute of Optronics, System Technologies and Image Exploitation (IOSB). Professors such as Heizmann, along with the alumni of their respective universities that work in the machine vision industry, are prime examples of the existing links between academia and industry. Universities maintain both formal and informal contact with their alumni and, in doing so, gain a valuable contact within the industry. ‘We have a personal contact with our alumni on at least a yearly basis, but if we know that a certain person works at a company with an interesting work topic, we keep in direct contact,’ explained Heizmann.
According to Heizmann, having this personal communication offers a variety of benefits. Universities are able to approach their alumni about potential collaboration projects or projects funded by the EU, for instance.
This contact can work both ways. ‘Our graduates also ask us if we are interested in a certain project when they know that this research topic is a specialty in our institute,’ said Heizmann. Forming collaboration is easier when personal contacts are known in the industry. ‘Graduates are people we know, we know what they do, what they are working on and that means a direct contact is always better,’ Heizmann confirmed.
Often the existing links between academia and industry are a direct result of the past experiences of the people involved. ‘Many leading academic scientists have a background in industry or institutes for applied science,’ commented Heizmann. Through maintaining these ties with former colleagues and alumni, a bridge is built between the two sectors that provides advantages to both sides. ‘Many interesting research topics come from industry,’ continued Heizmann, who spoke from an academic point of view. ‘The interesting topics come from the applications, and that means they come from industry. The contact with industry influences our direction of research.’
In return, the industry looks to academia for advice and information on the topics researched at the institutes. ‘We usually have the impression that academics work for industry, that we supply the manpower, the knowledge, the methods and the theory for industry,’ said Heizmann. He continued by explaining that, in most cases, industry will come to academia for questions about the use of certain methods or algorithms for a particular application. Without this relationship with academia, more resources in industry would have to be diverted into further research and information.
Certain projects that are funded by either the government or the European Union require the cooperation of both industry and academia to function effectively, which emphasises the need for collaboration between the two. The support from the academic side for these funded projects differs slightly from the normal advice it provides to industry.
European funded projects tend to have a broader, larger area of research or theory, and it’s often the case that academia looks to see if a certain methodology can be applied to a certain area of application. ‘In such a project we usually divide the work into parts,’ said Heizmann. ‘The first part, the theory part, involves, for example, how much the algorithms need to be adopted and made faster and more efficient to make them applicable to the topic that the industry is interested in.’
The industry side of the collaboration is expected to carry out any data collecting and monitoring of the processes being researched, which are often conducted on industrial sites. ‘They make the inspection hardware, they look for the cameras, the illumination and for the right setup, and they give us the images,’ Heizmann explained. ‘We then give academic feedback on what should be improved or what should be changed, and so the theoretical part is academic and the more practical part is in industry.’
Most of the collaborative projects that are undertaken at Fraunhofer IOSB are projects of applied research. ‘It is a defining characteristic of Fraunhofer institutes to have tight connections to industrial enterprises,’ said Heizmann. The Fraunhofer model of financing these applied science institutes requires that approximately one third of the entire budget is obtained by direct assignment from industry, while another third is through government-financed projects.
Forming collaborations with industry is therefore a vital part of Fraunhofer conducting its applied science research. In contrast to the Fraunhofer institutes, other institutes such as Max Plank and Helmholtz Centres concentrate on basic research. The research conducted at these sites is less directly involved in the vision industry, if at all, so collaborations are not made with the industry at these institutes to this extent.
To promote the idea of collaboration, events and activities take place throughout the year where scientists from academia and industry meet and share their experience, such as the EMVA business conference and the working groups of VDI – the association of German engineers. ‘I think the EMVA business conference [in Edinburgh in June] was a great platform for exchange,’ said Heizmann. ‘I have one cooperation project which we are preparing at the moment, and it is a direct outcome of a new contact that I met at this conference.’
Events such as the EMVA business conference or its vision forum provide the necessary organisation and infrastructure for pre-scheduled bilateral meetings between academia and industry to take place. The VDI has working platforms and working groups that provide services to promote collaboration between academia and industry. ‘They produce VDI guidelines,’ said Heizmann. ‘They gather the knowledge on a certain topic. For example, there is a guideline (VDI/VDE/VDMA 2632) concerning industrial image processing.’ This particular guideline introduces indicators describing the classification capability of a machine vision system, which have not previously existed. In addition, the VDI creates further opportunities for members of industry and academia to meet. ‘These working groups are also platforms to get into personal contact with people from different enterprises, branches and applications,’ explained Heizmann.
Through these platforms, members of both sectors can form collaborations. However, there are still opportunities for further development of the links between them. ‘I think we need more incentives to conduct cooperation projects,’ commented Heizmann. Many government-financed projects are not easy to obtain and require a large administration overhead, with EU projects specifically being the most complex. In addition, the funding is public tax money, and must therefore be carefully spent. Having further incentive to start collaborative projects will encourage both academia and industry to overcome these difficulties together and proceed with the expertise of both sectors.
Although platforms such as the EMVA events and VDI working groups exist, increasing the number of opportunities for the two sectors to meet will result in a larger number of collaborative projects taking place. ‘It would be good to have more platforms where persons from academias and industry can meet,’ agreed Heizmann. ‘These events don’t need to be very elaborate or expensive for the participants. Sometimes it is helpful just to get an insight into what others are working on.’ Ties between industry and academia could therefore be improved significantly simply by increasing the number of points of contact between the two sectors.
According to Heizmann, opportunities for improvement also exist within academia itself for creating ties with industry. ‘It would be a good idea to promote bachelor or master theses that deal with actual problems from industry,’ he said. ‘At the moment, it is not easy for me to offer such theses, because the right scientific supervision cannot always be ensured within an enterprise, for example.’ Because of this, many interesting and industrially relevant problems are not approached in academia. Promoting such industry-based theses will also encourage a greater number of graduates going on to work within the industry, which will increase the number of ties between the two sectors.
The European Machine Vision Association
The EMVA is a not-for-profit and non-commercial association representing the machine vision industry in Europe. The association was founded in 2003 in Barcelona by industry representatives from all over Europe as a network to promote the development and use of machine vision technology.
Ever since its foundation, it has been one of the top priorities for the EMVA to provide an interface between the machine vision industry and the relevant academic sector. Therefore, as a rather unique approach for an industry association, EMVA membership is explicitly open to universities and research institutes.
Through its actions, the EMVA is continuously working to address the improvements that could be made to ties between academia and industry, such as those previously mentioned by Heizmann.
In its ambition to support interaction between industry and academia on a European level, the EMVA has set up the annual european machine vision forum. This two-day event took place for the first time in September, and aims to foster interaction between the two sectors. It offers a platform for the sectors to learn from each other, discuss the latest research results as well as application-specific challenges and solutions, learn about emerging application fields, and to discuss research cooperation between industry and academic institutes. ‘The overall aim is to accelerate innovation by translating new research results faster into practice,’ said Prof Dr Bernd Jähne, coordinating director of the Heidelberg Collaboratory for Image Processing and an EMVA board member. ‘Many conferences target either academic researchers or applied industrial developers. However, on the European level, there is no forum where research and industry can meet and network. Thus, the European machine vision forum fills an important gap, building on the still continuing 20-year success story of the Heidelberger Bildverarbeitungsforum in Germany’.
The University of Applied Sciences Darmstadt currently offers German machine vision bachelor and master programmes. Building on this, the EMVA is jointly looking into setting up a European master course involving several European partners from academia and industry. With the strongly increasing use of vision technologies in all types of industry – from manufacturing via autonomous driving, healthcare and education to retail and logistics – the demand for well-educated specialists is still growing. The association plays an important role in channelling and communicating industry’s needs for the education of both researchers and practitioners to the relevant institutions.
