Kivi’s algorithm is plain and simple: every second Friday of the month, together with Mikrocentrum, they invite an engineering lecturer to a lecture on their field, followed by the opportunity to meet other engineers. Today was the floor at the AI Innovation Center at the high-tech campus Eindhoven for Albert van Breemen, CEO and CTO of VBTI, an AI engineering company that develops deep learning solutions for companies in agriculture and manufacturing. Van Breemen, whose company recently received a Gerard & Anton Award, took his audience on the hidden tracks of artificial intelligence and deep learning.
VBTI has successfully applied deep learning technology to agricultural robots and crop prediction systems. This required the development of a dedicated platform to make deep learning operational: AutoDL. The platform has automated many of the lifecycle tasks of deep learning development; With the support of VDL, VBTI is now taking the technology to a new level.
VBTI introduces robots in agriculture equipped with intelligent camera technology
His work is about “making automation intelligent,” says Van Breemen at the beginning of his presentation. “We want to support industries like agriculture, manufacturing, logistics and robotics in their transformation processes with deep learning and computer vision.” But first, what is deep learning?
Van Breemen presents a timeline showing three significant periods. “Not many people are aware of this, but artificial intelligence was mastered as early as the 1950s by creating machines that could perceive, reason, act and adapt. In the 1980s we had the second wave called machine learning. It was the age of algorithms that used data to improve their performance. Neural networks were created. Only after 2005 can we speak of deep learning: we started training deep neural networks with big data.”
Big data is crucial for this: the more data, the more deep learning options. But it’s not as if people are unemployed because of this development, says Van Breemen. “Most importantly, we need people to collect and select the data and comment on all of this so the machine can actually learn from it. After these processes, model training can begin, and finally it is time for deployment.”
leaf cutting robot
Consumer successes of deep learning can be found in autonomous driving, GO and chess successes in GO and chess or smart assistants like Siri or Alexa. “And now it’s time for the industry to get into deep learning.” VBTI/VDL are doing this by using AI to develop a robot to cut cucumber leaves. “It’s a really complex world because no cucumber leaf or stalk is the same, and yet the machine has to recognize them accurately. All these variations make it difficult, the deep learning toolbox can make this process robust.”
Van Breemen and his team have been working on the technology to support the defoliation robot for years. “Achieving 80 percent accuracy is easy, but the last 20 percent is extremely difficult. You always wonder what – and how much – data to collect and annotate, how to handle storage and versioning, and how to determine the quality of the data.” Van Breemen says he is happy and proud of the result that led to an effective robot, but he also knows that this can never be the end. “You never stop learning. You’re constantly collecting new data – more data means more deep learning capacity.”
