American researchers have developed a robotic system with artificial intelligence to help the selective cultivation of corn. Equipped with cameras, it is able to measure the inclination of the leaves.
To create a new, more productive maize variety, it is not enough to measure the cobs. The scientists scrutinize all the elements of the plant, in particular the angle of the leaves in relation to the stem. A plant whose upper leaves are vertical, and whose following gradually incline until they become horizontal at the bottom, will have optimal exposure to lightlight of soleilsoleil for the photosynthesisphotosynthesis.
To eliminate the need to perform this task by hand with a protractor, researchers from North Carolina and Iowa State Universities in the United States have developed a robotic system combined with adeep learningdeep learning dubbed AngleNet. They presented their findings in the journal Journal of Field Robotics.
A robot equipped with four pairs of cameras
The system uses a robot called PhenoBot 3.0, designed to pass between crop rows spaced 0.76 meters apart. It is equipped with four PhenoStero modules, each of which is composed of lights and two cameras to obtain stereoscopic images in order to allow the modelizationmodelization 3D. The cameras are placed at different heights to take pictures of leaves on different parts of the seedlings. butbut. The snapshots are then processed by AngleNet, an artificial intelligence based on a convolutional neural network. This is able to model the leaves and then calculate the angle.
The researchers compared the measurements obtained with AngleNet to manual measurements, and found a difference of five degrees, well below the acceptable margin of error. It should be noted that this technology is not intended to be deployed among all farmers. It is specifically intended for scientists working on the breeding of maize.
