The smart robot for collecting of apples

Main articles:

• Robots (robotics)
• Service robots
• Neuronets (neural networks)
• Artificial intelligence in agriculture (HSC)
• IT in agro-industrial complex in the world
• IT in agro-industrial complex of Russia

2020: Creation of the smart robot for collecting of apples using Microsoft Azure

On November 24, 2020 Microsoft announced that scientific group of department of data analysis and machine learning of Financial University under the Government of the Russian Federation and department of intellectualization, automation and robotizations of agricultural production of the Federal scientific agroengineering center VIM in technology partnership with Microsoft created the smart robot for harvesting of apples. Pilot tests of development will be passed in the gardening enterprises of Russia in the spring of 2021.

The neural network algorithm of the robot in which development scientists of the Financial university were engaged is capable to detect more than 97% and to collect up to 90% of fruits, at the same time a share of "false operations" when a system takes a background object for apple, makes only 3.5%. These indicators are much higher, than at other robots of the same appointment: the known prototypes detect on average 85% of fruits, and collect 75%. For training of neuronets Microsoft Azure cloud was used that allowed to accelerate considerably this process and also to reduce development cost in comparison with use of local capacities.

Quality improvement of detection of fruits became possible thanks to progressive algorithms of artificial intelligence, in particular, of deep ultraprecise neural networks which combine capabilities to object recognition on color, texture and a form, – Vladimir Solovyov, the Head of Department of data analysis and machine learning of Financial University under the Government of the Russian Federation emphasized. – The reliable and safe cloud of Microsoft Azure not only made possible implementation of this project in principle, but also helped us to achieve impressive results of everything in one and a half months.

Gardening – one of the least tsifrovizirovanny industries of agriculture: harvesting of the majority of fruit crops is usually made manually with involvement of the seasonal workers occupied with hard manual labor, at the same time up to 40% of fruits remain not collected. Use of the smart robot from first year will allow to increase by 30% revenues of farms due to reduction of a shortage of a harvest and also to solve a problem of shortage of human resources – Igor Smirnov, the head of department of intellectualization, automation and robotization of agricultural production of FNATs VIM noted.

The robot is intended for work in intensive gardens with height of kroner of 1.5 - 2 m. He collects fruits, since an upper tier, by means of the manipulators equipped with the captures created by specialists of the Federal scientific agroengineering center VIM. The average time of collecting of one fruit is 10 seconds, for an hour it can collect up to 288 kilograms.

Microsoft in Russia works hard contributing to the development of domestic science. In this project our cloud Microsoft Azure opens almost boundless opportunities before scientists and practicians, – Elena Slivko-Kolchik the head on work with the organizations of science and education Microsoft in Russia emphasized. – The smart robot created by the Financial university and the Federal scientific agroengineering center VIM is not only scientific development, but also the solution which will provide opportunities and competitive advantage to the gardening companies, having made them more effective.

The planned price of the robot is on average seven times lower, than at foreign analogs, and the payback for the European farms is about a year. For the Russian gardeners special price offers will work. In the spring of 2021 will pass pilot tests in the largest apple orchards of Russia. Then creators are going to bring the robot to the European market. Further similar algorithms will be developed for harvesting of pears and tomatoes. Besides, the possibility of use of the device for monitoring of productivity and recognition of the main diseases of cultures is considered.