"We stuck a toe in the waters of Artificial Intelligence and it yielded interesting results. We were looking for the value of AI in complicated things, while it turned out to be in very simple things."
Dennis Hoeks is Product Manager at Vencomatic Group, responsible for the Meggsius systems, among other things. In 2020, trainee Andre Hulsman did some research for him into the possibilities of AI, which inspired Dennis a lot: "The challenge is clear: systems should not only generate data, our challenge for the coming years is to link these data. And the next question is of course what we can do with that information."
"Artificial Intelligence is a huge hype right now. At Vencomatic Group we keep our heads cool and try to translate the possibilities of AI into practical applications. We looked at the data we have, and what we could learn from it. Our intern Andre Hulsman did extensive research on this in 2020."
"We started with data on where eggs are laid in the barn. We asked Andre whether we could predict whether the new eggs would be there again tomorrow, based on history. It is of course interesting for the farmer to see if the number of eggs in one place is going in the wrong direction. He also wants to see how he can get the eggs out of the house in the best possible way."
"In the end it turned out that chickens are really 'creatures of habit', haha. Chickens often lay their eggs in the same place. Andre did a statistical analysis with pattern recognition, and he also used a neural network to train the system to predict."
"The results were interesting, but predicting well proved difficult. The neural network certainly had the potential to be a lot more accurate, compared to the traditional statistical method. However, the quality of the datasets turned out to be insufficient and too small, so we didn't get a usable model directly but we did get interesting by-products. Such as the possibilities of visualising the locations of eggs in a 'heat map', which we are now rolling out in the Meggsius systems. We didn't expect to be able to do that already, apparently you can often do interesting things with a few simple steps."
"Our second lesson is that with AI you must first of all ask yourself what you are going to do with it. That is the challenge: applying the technology in the right way. A farmer will not let AI make decisions just like that, but as a tool it can help to make systems self-learning, for example. At Vencomatic Group we always talk about the 'autonomous barn'. In that case, decision-making will also have to be automated to a large extent. We are now taking steps in that direction."
"In the short term, the application of AI will probably remain hidden a little deeper in the technology, for example interpreting raw sensor data. This can give more insight for the poultry farmer, but in the future we also see possibilities in the field of management. For now, we will mainly use AI to translate the eyes, ears and nose of the poultry farmer into smart sensors. When we get more and qualitative data from this, there will be more and more possibilities to apply AI to this as well."
"In any case, with this assignment we have learned that the technology is ready, but that the challenge is mainly in our organisation itself. Of course, the farmers and the rest of the sector must also embrace this technology. Until then, we are certainly going to learn a lot from the data that we can get from the barn today and all the data that will be added in the near future."