How did you come to be part of the imaging/machine vision industry?
My first involvement in imaging technology was during my PhD at CSEM Zurich. I worked on high dynamic range CMOS image sensors, and had developed the patent for combined linear logarithmic response, which we branded LinLog. The main focus for this was automotive on-vehicle cameras but, as we did not understand the automotive supply chain too well at the time, we signed up for the Vision show in Stuttgart under a very preliminary concept to spin off this technology from CSEM. We called our project ‘Mountain View’, in strong analogy to the boom town of the dotcom age in Silicon Valley.
As part of Vision we entered the Vision Award – and ended up winning. The trade show was a huge success. We had never expected such a response, not so much from the automotive industry but from the industrial machine vision market. So, returning from Stuttgart, I focused on writing a business plan and a few months later we spun off Photonfocus. It turned out that we did not understand the supply chain of the machine vision market very well either at that time; at least, the combination of sensor design and industrial cameras did not work out well commercially. In 2003 I left Photonfocus and founded Awaiba, which concentrates solely on sensor design and manufacture.
How do you convince customers that they need machine vision?
For the industrial vision segment of our business, our customers are typically already well established in the machine vision market, as they either produce cameras or are larger OEM manufacturers of equipment embedding vision. So we do not need to convince them MV is a good thing. We sometimes have to educate them about the physical limits of opto electronics, but most are aware of the basic relationship between photon shot noise and SNR. In the non-industrial segment, mainly medical imaging, we sometimes indicate to our customers that machine vision exists and that many of their tasks could be done – or at least supported – by automated image processing.
What role does Europe have in the development of machine vision?
As more manufacturing moves to the Far East, the European MV industry has to remain innovative in terms of new technologies and applications in order to survive. Europe, therefore, is condemned to be an innovation leader, and will have to focus on new markets and applications otherwise we will not be able to survive in the long term.
Image sensor production is one example of this. Since there is virtually no production left in Europe for large-scale consumer electronics, sensor design and manufacturing companies have had to focus on alternative markets such as industrial vision. If we look at CMOS image sensors designed for industrial vision, probably 80 per cent of them come from Europe these days, at least in design. This is quite clearly a consequence of the European sensor companies losing access to the large-volume consumer electronics markets, which have moved almost entirely to Asia. I see a similar trend with machine builders, and vision suppliers should prepare themselves for machine building to follow production sooner or later. I don’t think it’s a good long-term strategy for our continent to move all large-scale production to the Far East, but we have to face this reality and live with it.
What are likely major growth sectors?
I see machine vision more as a technology than as an industry. The application of this technology is finding more and more areas of use. Some applications will be realised by traditional MV companies and others completely outside the industry, possibly through obtaining MV knowledge by acquisition. More close to traditional industrial machine vision, I see a large growth in industrial sensors, which embed a complete machine vision system but do not need the user to know about it. This could be devices such as light barriers, or size or volume sensors for the factory floor. Other growth areas for MV technology are in non-industrial sectors, such as home automation or medical assistance. Rich nations are getting older, and families are getting smaller. In the future, we will need more automation in our daily lives and in supervision during our old age.
What do you see as the most important technological challenges facing the industry?
One of the biggest technological challenges in the long term will be to adapt and incorporate successfully technologies developed for non-industrial applications – which may, due to the larger markets, outpace traditional MV solutions in terms of performance and price. At some point it may be cheaper and more flexible to buy a mobile phone with a camera and open source OS instead of a smart camera – these are already quite useful as data matrix readers.
In the near future, I see the transition to more distributed intelligence. Current applications require more and more resolution and image sensors produce this data at ever increasing speeds. Therefore, data processing has to move closer to the sensors, as bandwidth for data transmission cannot increase indefinitely.
What do you see as being the most significant commercial change in the industry during the years ahead?
We will see a lot of concentration in the component manufacture segment. This has started, but will continue strongly. Due to this, we will see fewer component vendors serving large-volume applications. On the other hand this trend will give opportunities for smaller players to focus on new niche markets and possibly non-traditional applications. So, at trade fairs, I expect to see on the one hand fewer larger companies, while on the other a more colourful spectrum of smaller players addressing more diverse applications and technologies. I also expect more Asian players in the vision component and system sector starting to export to Europe and USA, as they move up the food chain.
