Improving Agriculture with AI
Greenhouse gas emissions from the agricultural sector are estimated by the UN Food and Agricultural Organization to be about 14 and 18 percent of all polluting emissions. To put that percentage into perspective, the total global emissions from transportation sector is estimated to be around 13.5%. Of this percentage about half is due to plant cultivation which produces CH4 and N2O emissions. This is due to the current methods of irrigation, tillage and soil management, use of manure, deforestation, and use of pesticides. In the article Why is Soil So Important? The World Future Council notes that farming is the greatest contributor to soil erosion and is a serious cause for ecological concern without more sustainable practices. Additionally, habitats for native species are destroyed and fresh water reserves are depleted with the current practices to meet demand. Emerging Artificial Intelligent technologies seek to remedy inefficient current practices by boosting yield, controlling pests, and using less resources for growing crops.
Current Technologies
Irrigation
Currently, agriculture is the largest user of freshwater in the world. The artificial agent ConserWater predicts the appropriate amount of water a crop needs at anytime. It does this by using NASA satellite data, weather data, and ground soil moisture sensors with geospatial deep learning. This allows for precise watering of every crop reducing the use of freshwater used from overwatering and avoiding loss of crops from under-watering. In 21st century homestead: Sustainable agriculture III: Agricultural practices, Henkel states that “It is envisaged that with good management and knowledge of the optimum quantity of water needed for irrigation, the efficiency of even the surface irrigation method can be driven up to as high as 90%.” A.I. use in irrigation would allow for precise irrigation of crops on a plant by plant basis greatly reducing the overuse of freshwater for farming and lowering the impact from drought events.
Weed and Pest Control
Using computer vision and A.I. the “See and Spray” model, acquired by John Deere, is able to detect and identify crops and weeds then make a decision based on those results. This results in a 90% reduction in the use of herbicides according to Blue River Technology-the creators of the model-due to the pinpoint precision of the agent. The overuse of herbicides can cause pollution of soil and water due to the long period of time it remains active in the environment and can result in herbicide resistant weeds. This technology seeks to only target a weed directly instead of mass spraying and to differentiate between types of weeds to take the best course of action depending on the species. Currently the technology is only being used on cotton crops but the vision is to implement the software on a wider array of crops and include fertilizer, fungicides, and pesticides to target application to plants, fungus, and pests in order to use materials more efficiently and decrease the environment impact.
Detecting Disease
In order to boost yield and make the most of the land already used-as opposed to resulting to deforestation to expand fields- a mobile app known as Plantix is able to detect plant disease using a disease library and determine the best method of action to take and then recommends solutions. This monitoring would allow for the early detection of disease and pests and provide appropriate actions to take to minimize the loss of crops.
Future Outlook
Pollination
With bee populations in decline and favorable growing areas for crops such as vanilla in Madagascar not having a native species that can pollinate, hand pollination has become almost a requirement. Drastically inefficient, hand pollination requires the transfer of pollen from one plant to another manually requiring many man hours. Additionally, certain crops in favorable conditions, only have a low pollination rate. A.I. pollination agents could resolve this issue and ensure greater pollination rate in crops to produce greater yield per plant. Research is currently being conducted by researchers at at Japan’s National Institute of Advanced Industrial Science and Technology to create robotic bees to assist in cross pollination and Walmart has filed for a patent bee drones.
Greenhouses
The use of greenhouses provide a more controlled environment and have proven to be more efficient than outdoor growing and far more sustainable. High yields can be achieved with the ability to control the climate, irrigation, sunlight, insects, and soil content. The Netherlands is one country which already produces 35 percent of their vegetables in doors, taking up less than 1 percent its farmland. The controlled environment provides an optimal setting for A.I. agents with companies such as Microsoft and Intel participating in research in autonomous greenhouses. This research could lead to hyper efficient indoor farms almost fully controlled by A.I. agents to control all factors of growth for a crop resulting in farming which takes up less space, uses less fertilizer, water, pesticides, and herbicides while producing a greater yield.
Concluding Remarks
Global warming contributions from farming can be alleviated by standardizing the use of artificial intelligent agents in farming. Current applications of A.I. in agriculture and research have already proven in select cases that A.I. agents can assist in creating more food from less material-greatly reducing waste and environmentally detrimental chemicals in the agricultural sector.
