It might be hard to see the relation between high performance computing (HPC) and livestock farming at first. Nonetheless, four of the nine demonstrators in the CYBELE project concern precision livestock farming, actually two about pig farming, one about open sea fishing and one about aquaculture (about which you will learn more in upcoming blog articles). Moreover, it would not be difficult to find applications in dairy and poultry farms or even more extensive systems like sheep and beef farms.
In a previous blog article, we already learned how High Performance Computing can help a pork processing plant to get the right part with the right quality to the right customer. In addition, it enables them to learn more about what influences carcass and meat quality. Today’s slaughtering and processing plants in Europe are often large-scale operations with a lot of pig carcasses being processed daily in a highly-automated environment. So this is definitely leading to big data and, with the right traceability solutions like carcass tracking systems, to many possibilities for analysis.
But what about on the pig farm (or dairy or beef etc.) itself? Often these are still small-scale businesses and family farms, sometimes joined together in cooperatives, but most of the times one farmer is the manager of one or a couple of farm sites. A farmer is already a jack of many trades, to run a successful farm she or he needs to know about economics, market mechanisms, animal care, health, welfare, physiology, environmental impacts, nutrition, fertility, logistics and planning, people management, technology (feed and ventilation systems), etc. So why would we want to add high performance computing to that list?
Well, many farms are already big data producers on their own. And this will continue to grow in the future. Since we are dealing with such complex processes containing living animals, it is crucial that we are on top of everything that is going on at the farm to ensure on the one hand that pig health, welfare and environmental impact are secured and on the other hand that the farm is running a profitable business. To help with that, precision livestock farming tools come in handy. Whether it is something as frequently implemented as mechanical ventilation or feed delivery systems or novel solutions such as automated camera or RFID (radio frequency identification) monitoring systems, these sensors and systems all generate data. Besides the sensors, a general pig farm also has a lot of information on productivity (fertility, number of piglets born, slaughterhouse results) and costs (feed supply, medicine usage, water and electricity bills). This means that there are many opportunities for artificial intelligence, for example to identify risk factors, predict issues like a disease outbreak or optimize the decision making. In the CYBELE project we are working on several of these opportunities, like identifying risk factors for boar taint and developing warning systems for individual pigs’ health issues via RFID. And then with farm sizes of on average 2000 pigs in Flanders and thousands of such farms, HPC can definitely be needed.
That is what the outcome of CYBELE will be exactly, on the one hand developing the infrastructure needed for people that are not software engineers or ICT’ers but data analysts or advisors to be able to run algorithms on the HPC and to create impactful solutions for the agriculture and livestock sector. And on the other hand, demonstrating these solutions so that the farmers in the end can benefit from the HPC and leverage the data and information that is already or possibly available on their farm.