About 10-12 thousand years ago, man first began to engage in agriculture. From the 19th to the 20th century, it moved from individual use to industrial production, which is called agroindustry. Since then, the Earth’s population has grown significantly, which requires a new approach to ensure a sufficient level of yield. The development of AI has provided new opportunities to improve the agroindustry, thanks to computer vision, drones, robots, and other means. We will consider modern approaches becoming increasingly common and gaining significant popularity soon.
Vertical farming systems, also known as “Towers of Babel,” are a method of growing plants in multi-level structures, usually in buildings or special towers, instead of the traditional placement on a plot of land. The name comes from the image of the ancient Tower of Babel, because farms are built as tall multi-level structures resembling towers.
These systems focus on various technologies to achieve peak yields, evenly distribute water and fertilizers, and avoid plant diseases.
- Artificial Intelligence (AI)
- Predict plant growth, optimize watering, and fertilization.
- Detects plant diseases or stress through image analysis and sensor data.
- Sensors and IoT
- Monitor soil moisture, temperature, light levels, CO₂, and nutrients.
- Real-time data transmission for system automation.
- LED lighting with adjustable spectrum
- Automatically adjust the optimal light spectrum for different phases of plant growth.
- Hydroponics, aeroponics, and aquaponics systems
- Plants grow without soil in nutrient solutions or mist, which saves water and nutrients.
- Data analytics and cloud platforms
- Collect and analyze extensive data sets to optimize productivity.
This method is starting to gain popularity due to urbanization and climate change. Grand View Research estimates the size of the global vertical farming market in 2024 to be $8.15 billion, with a projected growth of $24.95 billion by 2030.
This allows various fruits and vegetables to be grown in absolutely different locations. Since vertical farming uses little space, it is a popular and preferred method for rooftop and other urban forms of agriculture. A prime example is Plenty, which combines AI, robots, and controlled environments to grow crops using less land and water.
A NatureServe study found that 22.6% of pollinator species, including bees, butterflies, and ground beetles, are at higher risk of extinction. New factors are now driving population declines, including military conflicts, microplastics, light pollution, soil antibiotics, and the combined effects of pesticides.
All this reduces the bee population, which occupies an essential place in the ecosystem. Bees are one of the species that have collective intelligence and influence the quality and diversity of crops. Bees ensure the yield of fruits, vegetables, nuts, and oilseeds. Plants pollinated by bees form larger fruits, directly improving farmers’ profitability. Each season, pollination by honeybees, native bees, and flies brings billions of dollars (US) of economic value. Annual global food production, worth between $235 billion and $577 billion (US), depends on their contribution.
But the bee population declines annually. Between June 2024 and February 2025, honey beekeepers reported a 62% loss of bee colonies, the highest rate in the history of observations. This has affected the agricultural sectors the most, particularly almonds, which depend heavily on bees for pollination. This problem can be kept under control thanks to AI, and the bee population can be protected. Let’s consider the main technologies that help in this:
- Computer vision. They use cameras near the street entrance, video annotation, and object recognition algorithms. Variants in the number of bee departures and returns, abnormal behavior, and the presence of pests help. For example, the startup BeeHero uses artificial intelligence, computer vision, and the Internet of Things (IoT) to monitor and analyze the activity of bee colonies. The collected data is transmitted to a cloud platform for further analysis.
- Acoustic monitoring. Microphones record sounds in the middle of the street. Signal processing algorithms and deep learning detect changes in the colony’s “beeping.” This can be used to predict swarming, stress, disease, and the absence of a mother.
- IoT sensors and Big Data analytics. Temperature, humidity, CO₂, and weight sensors are embedded in the street. To determine overheating or hypothermia, predict honey harvest, and control the impact of pesticides.
- AI in bee genetics and breeding. DNA and phenotype analysis using ML algorithms to create disease-resistant populations. A study published in the Journal of Economic Entomology offers a new approach to bee breeding using artificial intelligence. The researchers developed a system that uses markers based on the vitellogenin protein, which interacts with the main factors of colony loss. The application of machine learning algorithms allows predicting the effects of genetic variants of Vg on molecular functions, such as lipid, zinc, and DNA binding. The approach enables selecting bees with increased resistance to stress and disease.
An example of helping bees is tomato pollination robots with computer vision, like Arugga. Their technology combines computer vision and pulsed air. How it works: The robots move between rows of plants, select cameras and AI for flowers ready for pollination, and then hit them with a short pulse of air, which releases pollen in the same way that bumblebees identify them. To do this, NVIDIA used a real-time video processing platform. Thanks to this, AI models were trained on their own dataset and used YOLOv4 to publish flowers.
Pests are one of the main threats to the agricultural industry. They can not only destroy crops but also carry various diseases. Aphids carry more than 200 different plant viruses. Beetles, bedbugs, and thrips damage plant tissues, where fungi and bacteria multiply.
Various pesticides are used to destroy pests. These are deposited in fruits and are harmful to the health of people, animals, and soil. An alternative to this is drones for releasing beneficial insects, which are part of biocontrol.
How it works:
- Drones are equipped with special containers with beneficial insects that destroy pests or pollinators.
- The software distributes insect points over fields, gardens, or greenhouses.
- Flight planning algorithms and GPS are used to cover the territory evenly.
What insects are released:
- Predatory insects:
- Ladybugs (Coccinellidae) against aphids.
- Ladybugs (Chrysopidae) against aphids and moth larvae.
- Parasitoids:
- Trichogramma against pest butterfly eggs.
- Pollinators, bees, and bumblebees in greenhouses.
Advantages:
- Less pesticides, cleaner products, and soils.
- Time-saving, the drone covers dozens of hectares per hour.
- Precision, release in the right places and at the right time.
- Sustainability, preservation of biodiversity, and natural ecosystems.
An example of such a system is UAV-IQ, which combines biotechnology with artificial intelligence and drone technologies. Artificial intelligence analyzes data and optimizes flight routes, and drones distribute beneficial insects on plantations.
A new class of “small but smart” robots is currently being actively developed. These robots are not replacements for large tractors but complements them. They perform specific tasks in the fields. Due to their mobility, autonomy, and precision, these mini-systems increase the efficiency of agricultural production.
An example is Solinftec Solix, a line of autonomous agricultural robots designed for inspection, spot spraying, and continuous field monitoring. These robots are used for large-scale food production. The robot is equipped with four solar panels that provide its energy supply, artificial intelligence, and sensors.
Due to this, these robots have several advantages for the agricultural industry:
- Reducing the cost of chemicals.
- Improving plant health.
- Preserving the environment.
- Increasing the efficiency of agricultural production.
Robotics today is experiencing one of the highest demands for high-quality data annotation. Keymakr creates and annotates a variety of data, as the basis for training and testing robotic systems. With our annotation, companies can bring smarter, safer, and more efficient solutions to market.
New technologies that appear every day can bring us closer to fully autonomous farms, where people will only control the process. According to forecasts, the digital agriculture market could grow to $51.3 billion by 2033, where the main driver of development will be automation, artificial intelligence, the Internet of things, and robotics.
