The agricultural sector, considered the backbone of the economy of developing countries, contributes to 4% of the global GDP and more than 25% of the GDP, for some countries. In India, agriculture accounts for 18 – 20% of the GDP and employs more than 50% of the Indian population. With the rising demand for food to feed the 8 billion world population that continues to grow, there is immense pressure to improve the effectiveness and efficiency of farming practices to ensure a good yield. In addition, robustness to climate change and environmental vagaries that largely impact agricultural yield has become a critical need, more so in recent years.
Artificial intelligence has transformed several industries such as healthcare, finance, transport and construction, and has the potential to aid the agricultural sector address the above challenges. In this blog, we will look at how AI can make farming smart and sustainable. As with any new technology, adoption of AI into agriculture has its own roadblocks. We will discuss the current status of AI penetration into the agricultural sector and see how it can shape its future.
How can AI make farming “smart” and “sustainable”?
Artificial intelligence brings automation, insights and precision to farming practices. Automation drives efficiency and saves labor, insights facilitate informed decision making and precision improves sustainability and reduces costs.
Farm Automation: Driverless tractors, agri-BOTs and drones
Agricultural robots (also called agriBOTS) automate slow and repetitive farming tasks such as weeding, harvesting, seeding and spraying and free the farmer to focus on decision making. Drones are used to capture aerial images of your farm to monitor crop health as well as soil conditions. These can also be used for field surveillance and monitoring damage to crops. Seeding drones are used for cover crops (crops grown to limit soil erosion) and spraying drones for aerial spraying of pesticides.
Fruit and vegetable harvesting robots are in the nascent stages of development. These robots need to estimate readiness-for-harvesting through AI assisted vision and have to be trained to pick the fruit or vegetable in the right manner using flexible-arm robots.
Hand weeding is a labour intensive farming task and hence farmers turned to chemical based weeding that is not an environment or health friendly option. With AI-driven automation, weeding is now automated through weeding robots that can locate and eliminate weeds accurately and in an environmentally friendly manner.
Autonomous tractors that are more common now, can be used to not only optimally till and plough the land, but also perform precision planting of seeds, monitor soil composition and quality and spray fertilizers.
Predictive and Responsive Insights
Local Weather Prediction and Forecasting
AI generates predictive insights by analyzing large volumes of historical weather data and provides early warnings of imminent natural hazards like droughts and floods. This allows farmers to plan their farming activity accordingly and minimize crop loss. In addition, AI systems can suggest optimal sowing and harvesting times and even the kind of crops that would give a good yield, based on climate and weather predictions.
Yield Prediction and Price Forecasting
Global food production and distribution largely depends on accurate crop yield prediction and timely import and export decisions. Crop yield is a complex phenomenon and is determined by the genotype of the crop, the soil, weather and water conditions, the environment and the interactions between all of these. Deep neural network based models have been developed to predict the crop yield with reasonable accuracy. AI based systems can also be used to forecast crop prices to enable farmers to reap higher profits.
Pest Identification and Crop Disease Diagnosis
With the advent of AI, mobile phones and IoT, farmers can easily access and tap the community’s knowledge on pests and crop diseases. Images of pests and crops captured by drones and even the mobile cameras can be fed into AI systems and chatbots to identify the pest, diagnose the disease and take necessary actions. An android app, Plantix, developed by PEAT GmBH, an AI startup, enables farmers to diagnose crop diseases through images captured on their mobile phones.
Soil Composition and Quality
AI-driven systems can analyze soil nutrients and quality and even recommend the right crops to grow based on the soil composition as well as predict the irrigation and fertilizer needs.
Precision Farming: Smart Irrigation, Intelligent Spraying and Soil Analysis
Precision farming is a farming technique that leverages sensor inputs, hardware and software technologies like artificial intelligence, GPS and GIS, to use resources efficiently and in precise amounts to increase the agricultural yield. For example, farms are divided into zones and the soil composition, quality and moisture content is analyzed for each zone. This information is then used to predict precise irrigation needs. AI-driven systems can analyze soil nutrients and quality and even recommend the right crops to grow based on the soil composition as well as predict fertiliser needs. Intelligent spraying devices are used to minimize the amount of pesticides and chemicals used. These techniques not only save cost and improve yields, but also conserve precious natural resources like water and are more eco-friendly.
AI Adoption in Agriculture and its Challenges
Penetration of AI into the agricultural sector has been slow but steady over the last few years, and this is projected to grow in the coming years. Quoting Forbes, the global spend on AI-powered farming is projected to triple to $15.3 billion by 2025. The Indian government has acknowledged the potential of AI in agriculture and has allocated ~INR 8K crores to the Indian Council of Agricultural Research (ICAR) last year, for developing new farming technologies, their field evaluations and capacity building of farmers for their adoption. Microsoft is providing agricultural, land and fertilizer advisory services to 175 farmers in India, improving their average yield per hectare by 30%.
However, in spite of these efforts, adoption of AI into agriculture is not bereft of challenges.
AI systems need massive amounts of data to train
Accuracy of predictions largely depends on the training efficacy and this requires large amounts of unbiased data. It is hard to collect temporal data over large agricultural areas. In addition, any kind of prediction, say prediction of agricultural yield, requires processing multi-temporal data from different sources like satellite images and weather data and these have data noise and anomalies due to cloud and atmospheric effects. Thus data collection and preparation for training is hard.
Lack of experience with new technologies and tools
Adoption of AI into farming techniques and tools has been slow to a large extent due to the lack of experience of farmers with new technologies. This lack of experience brings fear and distrust. To address this, solution providers should invest time and effort in training the farmers in these areas, and introduce changes in a more gradual manner.
Data ownership, privacy and security
With no clear regulations and policies on data ownership, use of AI in agriculture can raise legal issues. Further, farm owners may be subject to data leaks, privacy breaches and cyber attacks.
Gap between farmers and AI engineers
For successful application of AI in agriculture, farmers need to work hand in glove with the AI experts for optimal training of AI predictive systems. However, in reality, there is a big gap here that needs to be bridged.
High cost of change
AI-driven farming tools and solutions need to be affordable and where applicable, open source, for easier adoption on a larger scale.
Will AI drive the next Agricultural Revolution?
Artificial intelligence has triggered a technology revolution in age-old farming practices. AI enables farmers to cultivate in a smart and sustainable manner, reaping higher and better yields while using fewer resources. If adopted at scale, AI definitely has the potential to power the next agricultural revolution. For this, farmers, governments and AI experts need to work in tandem to ease AI adoption.