AI-Powered Robotic Arm Helps Harvest Apples in Russia
An ensemble of neural networks delivers more than 95 percent fruit recognition accuracy in rain, fog and darkness.
Nikita Andriyanov, an associate professor at the Department of Artificial Intelligence at the Faculty of Information Technology and Big Data Analysis at the Financial University under the Government of the Russian Federation, has developed a computer vision system for a robotic gardener. The project earned him an award in the Information and Communication Technologies category.
Andriyanov said his team set out to automate agricultural processes, particularly in sectors that rely on heavy manual labor. Harvesting is one of them. Researchers designed a robotic manipulator arm to pick apples in orchards.
High Accuracy With Room for Expansion
The key difference in his approach is the integration of several neural networks into a single system. This ensemble architecture pushed recognition accuracy above 95 percent. The AI can detect apples in virtually any conditions, including rain, fog and nighttime darkness. The neural network also generates an algorithm for harvesting and calculates the manipulator’s movements.
In the near future, Andriyanov plans to train the system to recognize tree diseases, determine when orchards require maintenance and communicate with other robots. The technology is already being deployed at agricultural complexes.
