Scientists at Moscow Polytechnic University are developing an intelligent control system for grain harvesters. It will allow automatically adjusting harvesting parameters to the condition of crops and significantly reduce crop losses. The research is supported by a grant within the framework of the "Priority 2030" program.
When harvesting wheat, grain losses can reach 6.6%, with more than half occurring during the cutting of the crop mass. The reason is the human factor: even an experienced machine operator cannot continuously monitor the machine's operation. Serial combines react to changing conditions with a delay of 3–6 seconds.
The development is based on a neural network computer vision model. The system processes data from cameras and a depth sensor on the header, evaluating plant height and density, lodging degree, and crop mass volume. Based on this, the combine's speed and the cutting mechanism's frequency are automatically adjusted. Parameters are updated several times per second.
The new approach reduces crop losses, minimizes grain damage, and prevents impurities from entering the bunker. The development will be a step towards creating fully autonomous agricultural machines.
