Artificial intelligence from Russia will help the plants to conquer space

© NASAКапуста of Mizuna growing aboard the ISS. Archival photoArtificial intelligence from Russia will help the plants to conquer space© NASAПодпишись to daily updates RIA Science

Thank you for subscribing

Please check your e-mail to confirm your subscription

Scientists from Skoltech created a system of machine learning, which will help space agencies of the world to find the «right» plants to provide for future long-duration expeditions to outer space with the necessary amount of biomass and oxygen. Their findings have been presented in the journal IEEE Pervasive Computing.

«The main advantage of our method is that a three-dimensional image is sufficient to obtain for each species of plants only once. After that, to predict the biomass growth, it is sufficient to use a simple camera. This greatly simplifies and reduces the cost of systems designated for the forecasting, control and optimization for greenhouses and artificial life support systems in space» – said Dmitry Shadrin, a graduate student at Skoltech, whose word brings the press service of the University.

Long-term flights into space today believe NASA and Roscosmos, will require the establishment of a fully Autonomous life support systems, allowing water, oxygen and all necessary nutrients for an unlimited period of time.

The key to their creation are now considered plants, and various unicellular algae, capable of producing biomass in large quantities and with great speed. Over the past two decades, scientists have progressed in this direction by creating on Board the ISS two greenhouses and they grow cabbage, lettuce, asters and many other plants.

Such successes are forcing biologists, space doctors and other researchers to think about how many plants need for survival of the crew flying to Mars or other planets. Their excess can do the mission impossible and prohibitively expensive, but it will doom future followers Mark Watney of «the Martian» to a slow death.

Despite the fact that scientists are studying plants for thousands of years, to prepare such assessments is not so easy, as their rate of growth and a set of biomass depends on many different biological and physical factors — the amount of moisture and micronutrients in the soil, light levels and dozens of other things. In addition, the biomass is quite difficult to «weigh» without killing the plant itself, that prevents the estimate of the speed of its growth.

Shadrin and his colleagues at the Institute after, Rupert, Gerzer (Rupert Gerzer), Tatiana Podladchikova, and Andrey Somov, find out how you can quickly and accurately enough to make such assessments, watching the growth of dwarf tomatoes with the help of three-dimensional and two-dimensional cameras.

Analyzing the state of the tomato at different phases of growth, the Russian scientists were able to infer some regularities associated with a set of biomass, and used them to create machine learning systems that can assess these characteristics by analyzing a simple two-dimensional pictures of the leaves of tomato and a three-dimensional model of a plant.

As shown by further monitoring, the program correctly predicted the growth rate of tomatoes and several varieties of lettuce, for the first 30 days of life after the landing. It allows to use it not only to render the «space» life support systems, but also to optimize the operation of greenhouses.