Plant on sight: Autonomous gardening robots will destroy weeds and slugs with a laser
- Статьи
- Science and technology
- Plant on sight: Autonomous gardening robots will destroy weeds and slugs with a laser
Russia is developing gardening robots that can detect pests and destroy them with a laser. Slugs, weeds, and insects no longer fall into the hands of an agronomist, but into the field of view of a neural network. Scientists have created a line of autonomous platforms capable of digitally monitoring fields, targeted pest control, and harvesting crops with little or no human involvement. From a technological point of view, such machines combine several rapidly developing areas at once: lidar navigation, machine vision and artificial intelligence. However, the mass implementation of such solutions depends not only on engineering capabilities, but also on how economically feasible they are for agricultural enterprises, industry experts interviewed by Izvestia emphasized.
Robotic platforms for garden and field
Scientists at the Federal Scientific Agroengineering Center VIM have developed a line of autonomous robots to optimize work in gardens, fields and greenhouses. Among them is a platform with a control system based on laser rangefinder. Such a robot gardener is able to inspect plantings and determine which diseases and pests affect certain crops. The machine scans the surrounding space using LiDAR technology and builds terrain maps based on the resulting point cloud.
— Such a platform can independently move around the garden in an unmanned mode and count apples and other fruits using a camera, as well as determine what the plants are sick of. It is a digital monitoring system that can be configured for different parameters. What we put into the neural network, it will determine," Igor Smirnov, head of the Department of Technologies and machines for horticulture, Viticulture and nursery Breeding at the Federal State Budgetary Scientific Institution FNAC VIM, Corresponding Member of the Russian Academy of Sciences, told Izvestia. — You can track various types of diseases and pests. It all depends on how we train the program. This is a tool that will allow you to collect information from the field, and then analyze it on a computer.

In addition, the modular design of the autonomous platform allows you to install various devices on it, including a special laser nozzle for pest control in real time.
— We have already conducted a preliminary experiment on using a laser installation to remove weeds. It can also be used to kill insects. For example, there is a well-known experience in China, where a laser was used to reduce the mosquito population. In our country, the slug problem is now widely known. They can also be dealt with using such an installation. We can train a neural network to recognize any pest. We also have algorithms for controlling the laser installation itself.: we will set her a goal in the form of a snail or a slug, and she will fight them," Igor Smirnov said.
According to the developers, slugs will be even easier to deal with than weeds, as they are better visible using machine vision, but it is necessary to conduct field tests of the installation against shellfish. This can be done next year.

The new robot gardener can also be supplemented with baskets or crates for harvesting. Another possible function is contour pruning of tree crowns.
A digital garden with robots
In addition to the farmland monitoring platform, scientists have also created a new hydropneumatic injector based on a robotic platform designed for in-ground irrigation and fertilization. Such a machine is capable of injecting nutrients in the form of hydrogels directly into the root zone of plants. These can be fruit, berry, vineyards, essential oil and ornamental plants, the authors of the technology told Izvestia.
Another development of the center is a complex for autonomous apple harvesting, which is equipped with a vision system based on a convolutional neural network, a frame manipulator with a telescopic guide and an interchangeable gripper. According to the scientists, the use of the complex makes it possible to reduce the complexity of harvesting fruit crops, solve the problem of a shortage of seasonal labor, and reduce production costs for harvesting by 20-30%.

It is important that all platforms operate on the principles of green energy.
— We are moving away from diesel or gas engines to hybrid technologies, and our robots can work 24/7 using our battery technology. It is environmentally friendly and economical," Alexander Podkovyrin, an engineer at the Laboratory of Machine Technologies for Cultivating and Harvesting Fruit and Berry Crops at the Federal Scientific Agroengineering Center VIM, told Izvestia.
From a technological point of view, such platforms combine several areas at once, which are developing especially rapidly: lidar navigation, machine vision and trainable neural networks. This is a typical way of developing robotics today: a universal "chassis" plus specialized modules and algorithms for a specific task, Leonid Drobyshevich, an NTI technology expert, told Izvestia.
— Key technological challenges are related to the accuracy and speed of recognition in real conditions. The neural network must confidently distinguish a slug from a clod of earth, and a weed from a cultivated plant under different lighting conditions, humidity and contamination of the chambers. The laser system needs to be aimed at the target in a fraction of a second and not damage neighboring plants. A separate task concerns autonomy: batteries and hybrid power circuits should ensure operation throughout the season without frequent recharging," the specialist said.

In this case, it is interesting that the system is able not only to find a problem, but also to respond to it immediately, Nikita Rumyantsev, founder and CEO of SkyFarm Ecosystem, an expert in the field of AgTech, commented on the development. For example, a robot can destroy a weed with a laser without affecting neighboring plants. Similar technologies are already being used in the world. In the future, the laser can also be used to control individual pests, such as slugs or caterpillars. However, tests will first be required to show how effective and safe this method is for the crops themselves.
— In my opinion, the main perspective of such robots is much broader than pest control. Today, the machine monitors the plantings, tomorrow it helps to process plants or harvest crops. And if you combine it with other systems, you can gradually automate the whole farm. This is exactly the approach we are developing at SkyFarm. In our modular farms, we combine sensors, cameras, climate control systems, lighting, irrigation and plant nutrition," he said.
It is important that the agronomist does not have to constantly check the equipment. Instead, he should see the state of the entire production, receive recommendations and manage processes through a single system. The economic component is equally important. The farmer is not interested in the number of neural networks in the robot, but in how much money the technology will save, which crop will help save and how quickly the investments will pay off, Nikita Rumyantsev emphasized. Therefore, the mass implementation of such solutions depends not only on the engineering capabilities, but also on their practical benefits for the economy.

The scientists presented their developments at the Harvest of Technologies Agricultural Festival, which was held as part of the Decade of Science and Technology in Russia.
Переведено сервисом «Яндекс Переводчик»