Machine farmers will work in the field

Identifying agricultural products that need to be harvested, spraying pesticides and harvesting are things that farmer robots can do in the field in the future, Israeli scientists said.

Today's robots are almost impossible to identify the harvestable fruits on trees. Photo: qwikstep.eu.

Farmers in developed countries have used tractor, combine harvester, cow milking machine and many other modern equipment. But picking fruits and vegetables is something that almost no machines can do. For example, the machine cannot determine which ripe grapes or which are still young. Each fruit, branch, leaf, flower has its own shape, color, size. Different lighting conditions also change the look of agricultural products. Robots can recognize yellow, red, or purple fruits among green leaves, but they will be "helpless" if the fruits are green like leaves.

For these reasons, programmers cannot create software to help robots or computers identify harvestable produce.

' The technology for making robots for agricultural operations has been around for a long time, but now we are starting to consider making them for sale. I think such robots will appear in the fields for the next five years , 'said Yael Edan, a mechanical and robotics expert at Israel's Negev University.

Discovery said that to help computers identify harvestable plants, flowers, fruits, tubers or leaves, Edan and his colleagues sought to build smart sensor systems. For example, they use multi-spectral cameras to analyze light wavelengths to objects. In the image that the camera recorded, the object appeared with a unique appearance in all lighting conditions.

Along with other types of sensors and computer programs, the team also wants to create a machine 'brain' to help robots learn from the mistakes of the operation process to improve the quality of work.

' We will create an algorithm to help robots see things as simple shapes. When the produce is partially obstructed by leaves, it will still recognize, 'Edan said.

Testing shows that robots built by the Edan group can determine the status (green, ripe or broken) of 80-85% of the fruits on the tree. Their goal is to raise that number to 90%.

After the robot identifies the agricultural product to be harvested, it takes the product with a tool capable of grasping. The biggest challenge of making this tool is that it must come to the right place and take a moderate amount of force so that the agricultural product does not crush, crush, or break.