Artificial Intelligence: a Potential Tool for Precision Agricultural Farming

The global food production must be increased to feed ever-growing population. Thus, the sustainability of agriculture field is the key to guarantee food security and hunger eradication. Moreover, weather and climate change conditions, together with the sustainable water management due to water scarcity, are crucial challenges. For these reasons, it is necessary to establisha strategic shift from the current paradigm of enhanced agricultural productivity to agricultural sustainability. To foresee efficient solutions, helping farmers and stakeholders to enhance their decision by adopting sustainable agriculture practices is a crucial choice, especially the use of digital technologies including Internet Of Things (IOT), Artificial Intelligence (AI), and cloud computing. AI is a creative tool that simulates the human intelligence and ability processes by machines, mainly computer systems, robotics, and digital equipment. The AI technique is being used in several sectors which are seeing the fastest growth in the recent years such as finance, healthcare, retail, pharmaceutical research, intelligent process automation, and marketing. Now-a-days, there is a bright prospect of AI in agriculture.  The most important goal of AI in agriculture is to provide precision and forecasting decision in order to improve the productivity with resource preservation.  Machine learning (ML) is one of the central themes of AI and helps people to work more creatively and efficiently.

Currently, the ML algorithms are mostly used in the main four stages (preproduction, production, processing, and distribution) of the agriculture supply chain. Different machine learning (ML) algorithms are used for crop yield prediction such as the Bayesian network, regression, decision tree, clustering, deep learning, and artificial neural network (ANN). The LS-SVM (least-squares support vector machine) method is used to predictsoil properties soil. The SaE (self-adpative evolutionary) ML algorithm used to boost the performance of the extreme learning machine (ELM) architecture to estimate daily soil temperature. Additionally, a novel method named the CSM (Crop Selection Method) to resolve crop selection problems. In fact, to achieve an effective irrigation system (better decision in when, where, and how much to irrigate), producers used soil moisture data, precipitation data, evaporation data, and weather forecasts as input data for simulation and optimization of predicted models based on ML adequate algorithms. In addition,ML algorithm associated with other technologies such as sensors, Zigbee, and Arduino microcontroller is efficient for prediction and tackles drought situations. Moreover, ANN feed-forward and back-propagation technologies are used to optimize the water resources in a smart farm.

A German-based tech start-up PEAT (Progressive Environmental and Agricultural Technologies) has developed an AI-based application called Plantix that can identify the nutrient deficiencies in soil including plant pests and diseases through using image recognition-based technology. The farmer can capture images of plants using smart phones. Similarly, Trace Genomics is another machine learning-based company that helps farmers to monitor soil and crop's health conditions.

The crop production stage is the second phase in the agriculture supply chain which is affected by numerous parameters. Among them are the weather forecasts (sunlight, rainfall, humidity, etc.), crop protection against biotic stress factors (weeds and pathogens) and abiotic stress factors (nutrient and water deficiency), crop quality management, and harvesting. Many different ML algorithms are used to simulate effective models for weather prediction (ANN, deep learning, decision tree, ensemble learning, and instance-based learning), for crop protection (clustering and regression, ANN, deep learning), weed detection (ANN, decision tree, deep learning, and instance-based learning), crop quality management (clustering and regression), and harvesting (deep neural networks, data mining techniques and ANN). ML algorithms are also used to predict the fruit ripening stages and fruit maturity through fruits color.

The processing stage is the third stage in the agriculture supply chain where to achieve a high quality and quantity of food product, food industries use modern food processing technologies by installing software algorithms based on ML (machine learning). In fact, support vector machine (SVM) classifier and artificial neural network (ANN) models are used to detect the presence of nitrosamine in the red meat. Additionally, differential scanning calorimetry combined with ML tools is used to determine the milk characteristics and authenticity and to detect fraud.

The distribution cluster is the final step in the agriculture supply chain which is the connection between food production-processing and the end user or final consumer. In transportation and storage steps, the mainly used algorithms are genetic algorithm, clustering, and regression. These predictive techniques aim to better preserve the food product quality, to ensure safe food products and to minimize the product damage by tracing the product. For the consumer analytics, ML techniques such as deep learning and ANN are used in the food retailing phase for predicting consumer demand, perception, and buying behavior. For the inventory management, the use of ML genetic algorithms helps in predicting daily demand and to ensure that there are no inventory-related problems.

There arenumerous AI-applied technologies in the agri-food sector; these are as robotics and mechatronics, drones, geographic information systems (GISs), blockchain (BC), and satellite guidance. Revolutionizing machines, often called "agribots" are now used in agriculture for all kinds of activities, namely, soil preparation, seed sowing, weed and pest treatment, irrigation, fertilization, and ultimately, grain and fruit harvesting, minimizing effort and energy cost. As a whole crop management, agricultural drones are now able to supply water, fertilizers, herbicides, and pesticides and even film, capture images, and generate maps in real time of plants and field in order to help farmers take management decision.

Today, farmers use drones for livestock surveillance for monitoring illnesses, injuries, and even pregnancies. Based on geospatial technology that relies on satellite, GIS is applied on several fields of agriculture: crop management, irrigation control, yield estimation, disease and weed control, farming automation, livestock monitoring, vegetation mapping, erosion, and land degradation forecast. Blockchain (BC) is another technology that answers to consumer's awareness about food origin, quality, and mainly, safety. BC affords transparency, trust, certification, and traceability of food product supply chain from farm to table, where every single operation and data are timely registered, saved, and secured, not in a single central server nor under a single control, but in a common platform database where every user could access and take part in transactions.

Agriculture is one of the most vital fields for humanity and plays an important role in the economic sector. The first products of agriculture are used as inputs in several multiactor distributed supply chains, including four stages of the agriculture supply chain (preproduction, production, processing, and distribution) in order to reach the end user or consumer. Due to several challenges in the future for the agriculture and various factors such as climate change, population growth, technological progress, and the state of natural resources, it is necessary to use the digital technologies at every stages of agriculture supply chain such as automation of farm machinery, use of sensors and remote satellite data, artificial intelligence, machine learning for improved monitoring of crops, and water, for agricultural food product traceability.

The applications of the AI and ML algorithms in different stages of the agriculture supply chain couldimprove agriculture and food industries.Interestingly, some public and private organizations in our country are now trying to develop or explore AI-based technology suitable for our agriculture. Therefore, the application of the advanced AI-based technologies would be a potential tools in the agriculture of Bangladesh in near future through sinking employee training costs, reducing the time needed to solve problems, reducing human intervention, and offering an automated good, accurate, and robust decision-making on the right time with low cost.


Dr Md Monirul Islam is Chief Instructor (Deputy Director), ATI, Homna. Attachment: YRFP project, Department of Agricultural Extension, Khamarbari, Dhaka