As part of Passion Sponsorship Program, engineers from Sii Poland are implementing the EVAA project based on Artificial Intelligence in Gdansk – Ecological Vertical Agriculture Assistant. The initiative, in which about a dozen engineers from the company is involved, is implemented in cooperation with experts from the Tri-City scientific community. EVAA has been recognized by Internet users in the international Innovate FPGA competition for attempting to solve problems of famine and ecological threats technologically. Paweł Czapiewski, Team Leader at the Embedded Competence Center and EVAA project leader talks about the project assumptions.
What is EVAA?
Paweł Czapiewski, Team Leader in Embedded CC: EVAA is an abbreviation of Ecological Vertical Agriculture Assistant. This is a project in which we try to teach artificial intelligence how to grow plants. EVAA is a system in which lettuce grows under the watchful eye of the machine. In our approach, lettuce is cultivated hydroponically. This is a special way of growing plants, which means that they are immersed directly in water from which they draw nutrients. This is an ecological way of growth – exploitation of surface and water is reduced as opposed to traditional soil crops.
We are talking about a research and development project. What is its experimental character?
P.Cz.: In our approach, we want to use reinforced learning to control the physical process of plant development. Reinforced learning is a special kind of artificial intelligence. The way it works is that the algorithm receives only a feedback signal in the form of a prize, if lettuce develops well. What is the optimal way to get things done? The algorithm must come up with its own solution. Together with the team and the machine, we try to achieve optimal results.
EVVA is a technological project that helps fight the problems of modern ecology.
P.Cz.: Yes, the mentioned ecological aspect is extremely important here. However, in essence, it is a technological project in which artificial intelligence plays a forefront role. Artificial Intelligence is an extremely hot topic nowadays. Its special kind, namely reinforced learning, became widely popular when the AlphaGo program won against a man playing Go. It is a much more difficult game for machines than chess, where the number of combinations of possible moves to choose from is much larger. We want to use a similar approach – instead of playing the game – to optimize lettuce farming.
What experts make up the team involved in this project?
P.Cz.: The project is carried out by Sii engineers as part of Passion Sponsorship Program – the team consists mainly of Embedded Competence Center’s engineers. They play a various roles: designers, constructors, programmers, testers, and sometimes carpenters, as even the wooden structure was made with their own hands. External experts are also involved: Piotr Wierzgała, an expert in machine learning from the Tri-City, and Tadeusz Puźniakowski, lecturer at the Polish-Japanese Academy of Information Technology.
The project is conducted as part of Passion Sponsorship Program. How does the participation in the Program help you?
P.Cz.: Without Passion Sponsorship Program, the EVAA project would not have been possible. This system requires a large amount of specific hardware. It was purchased from the Program’s funds.
At the same time, project participants can their make their passion tangible: EVVA is developed during off-work time when engineers are not involved in commercial projects. For each of us, this is a great opportunity to deepen our competence in technical areas or gain new skills.
What competences did the team gain and develop thanks to this project?
P.Cz.: As part of the conduct of this project, a comprehensive system was created that enables visualization, processing and collection of data, including camera images. It was not without processing and recognition of images to properly determine the reward signal for an intelligent agent. During the work on the system, many different sensors and actuators had to be integrated with each other. Our competences in each of these areas have increased significantly. Our knowledge of technologies and tools, such as programming in Python or C, Flask, Grafana, OpenCV, Blender, CI / CD, as well as testing embedded devices has also extended.
EVVA is a really fascinating experiment that we can jointly lead and watch as we progress. There is still some work ahead of us, but we are already happy with the first success of our solution. We won the Iron Award – an award in the prestigious Innovate FPGA competition organized by Intel. In addition to the prize, work with artificial intelligence and observation of how it develops certainly gives great satisfaction. I highly recommend participating in such experiments!
Get to know other projects of Sii employees within the Passion Sponsorship Program.