IoT-Based Robotic Systems Integrating Artificial Intelligence for Sustainable Crop Protection in Agriculture

Authors :

S.Pugazhenthi and S.Sasikumar

Address :

1Department of Computer Science, Sri Moogambigai Arts and Science College, Mallupatti, Dharmapuri

2Department of CA, karpagam Academy of Higher Education, Echanari post, Coimbatore

Abstract :

Advancements in technology, including the Internet of Things (IoT) and Artificial Intelligence (AI), greatly impact agriculture. The study investigates the statistical applications of AI-driven robotic devices based on the IoT as a basic emphasis. Using manual effort and chemical fertilizers, traditional farming can be highly attributed to inefficiency, health problems and environmental effects. The paper proposes an AgriBotIQ, a revolutionary platform that uses robotics based on IoT to monitor and analyse with accurate participation in plant management. Autonomous robots can collect information based on plants and their habitats by using imaging devices and sensors like soil moisture, humidity, temperature, and many others. Machine learning (ML) algorithms search the database for anomaly detection, threats, and crop trends. To identify the crops that are diseased or healthy, ML is integrated with computer vision. The suggested AgriBotIQ also eliminated weeds, boosting the output by neglecting unneeded waste and chemicals. The emerging IoTs have allowed better remote plant monitoring in more versatile and précised. Overall productivity and protection of crops are possible by statistical analysis and real-time notifications of the proactive decision-making outcomes. By combining IoT and AI, the future agricultural crop security will improve greatly.

1.Introduction

The UN projects that by 2050, the world's population will have risen to 9.7 billion, calling for a 70% increase in food production. The way forward for businesses in this situation is to find ways to boost agricultural output while reducing the negative effects of conventional farming on the environment. Overuse of chemical pesticides and fertilisers has eroded soil, contaminated water sources, and harmed the well-being of humans and animals. The study wants new ideas to boost agricultural productivity in a sustainable and environmentally friendly way [1]. Maximum production, optimal decision-making, effective use of resources, and environmental sustainability are all guaranteed by "smart farming" thanks to objective data obtained by sensors. The cornerstone of sustainable agriculture will be methodologies grounded in robotic solutions and artificial intelligence [2]. Improved decision-making is possible with the use of drones and satellites for remote sensing by closely monitoring soil health, plant energy, and various other environmental parameters [3]. Artificial Neural Networks, ANNs, and other contemporary AI problemsolving tools enable us to address unique agricultural issues. For farms, "smart farming" means implementing state-of-the-art technology like AI and the IoT, which completely changes how things are done [4]. The Internet of Things (IoT), cloud computing, and machine learning allow for all-encompassing environmental monitoring. By categorizing data, researchers may choose the optimal circumstances for the growth of different types of crops [5].

By monitoring several agricultural metrics, the Indian-based Fasal IoT platform assists farmers in making more informed decisions regarding their operations. An AI-powered engine powers the site. The application of AI and data analytics enables farmers to attain future crop control and optimum productivity [6]. It takes effort and time to attend to all the requisite logistics, such as collecting, sorting, shipping, and marketing crops. Better agro-business practices can be achieved using smart agriculture technologies to address and alleviate these challenges [7]. Precision farming is one sustainable option; it increases output with precisely determined inputs while decreasing the usage of inputs that could be damaging to the environment, such as pesticides [8]. To increase productivity, decrease input costs, and improve crop yields and quality per worker, a new method called "smart farming" is being implemented [9]. The study employs cutting-edge ICT techniques such as the Internet of Things or IoT global positioning systems (GPS), sensors, robotics, drones, actuators, precision machinery, and data analytics to ascertain the farmers' needs and put suitable solutions into action [10].

The grain yield mapping framework includes sensors for moisture, grain volume, GPS (global positioning system) antenna and receiver, and travel speed. Crop yield monitoring uses grain flow sensors, which use mass and volume flow methods. A yield sensor measures the stream rate every one to two seconds during collecting or the harvested amount per unit area [11]. With the help of autonomous robots, weeding becomes much more efficient, and less pesticide is used. Drones can be used to successfully spray pesticides and monitor crops without requiring an excessive amount of human labour. Crop readiness assessment and forecasting yields are useful tools to analyze and forecast the quantity of high-quality harvest available for sale [12]. Key contributions of the study include,

The remaining part of the study is organized as follows: Section 2 describes the various existing studies related to machine learning, IoT and AI technologies for smart farming; Section 3 describes the proposed AgriBotIQ approach of how it works on precision agriculture; Results and discussion based on improvement in the smart farming is given in section4, and the study concludes in section 5 with their future works.

2. Related work

Mesías-Ruiz et al. [13] proposed the importance of AI, ML, and other developing technologies in protecting crops from the effects of climate change and meeting the increasing need for food. The study showed how crop protection has progressed from its earliest forms to its current state and how ML algorithms have helped with precision. Although other creative methods may be disregarded, the article discussed emerging technologies such as intelligent sensors and AI-based robots for the next generation of crop security systems. Shahab et al. [14] proposed smart farming powered by the IoT and using AI and ML algorithms to help improve agricultural productivity sustainably, which in turn helps achieve the Sustainable development goals (SDGs) set out by the United Nations. The article focused on the most recent findings in smart farming, which emphasizes the capabilities of machines and the application of UAVs and robots in farming. It also stressed the significance of wireless communication technologies in managing sustainable agriculture.

Ali et al. [15] outlined state-of-the-art IoT and smart farming techniques to maintain sustainably produced crops in response to critical global challenges like climate change and rising food demand. These technologies made agriculture more efficient, productive, and cost-effective by keeping tabs on crops, controlling resources, and forecasting yields. Although the study provided potential benefits like reduced energy use and improved illness management, it can fail to resolve issues like high costs and a lack of technical knowledge. Javaid et al. [16] proposed that many agricultural jobs, such as crop cultivation and soil analysis, have become much easier to accomplish with recent breakthroughs in AI. With the help of AI, farmers can choose the best seeds, plant them at the right times, and apply nutrients so that their crops are ready for market with minimal waste of resources. Health monitoring systems are one method that allows farmers to see their crops in action right now. However, obstacles such as price and lack of accessibility to technology still stand in the way of broad AI implementation in farming.

Using a full-scale prototype made with IoT and 3D printing technologies, Catota et al. [17] proposed a solution to the issue of inefficient agricultural resource management. Incorporating 3D printing for sowing and the IoT Arduino Cloud for monitoring showcases effective cultivation by providing real-time visualization of factors. Despite showing efficient cultivation, limitations may include problems with scalability, cost-effectiveness, and the complexity of combining different systems. Riskiawan et al. [18] presented an IoT, AI, and LSTM-integrated automated greenhouse environmental control system. Effective climate monitoring and management is available. However, there may be drawbacks, such as difficult system integration, problems with scaling, and high costs. Some people doubt the system's usefulness in real-world scenarios because of its dependence on LSTM training data and the ongoing requirement to gather data.

Parasuraman et al. [19] studied cutting-edge smart agriculture and Internet of Things (IoT) technologies to combat pesticide overuse and agricultural diseases. In addition to potential problems with efficiency and scalability, integrating different types of automation might be difficult. To be prepared for actual use, the suggested IoT plant recognition and watering design needs more validation and testing. The authors of the work by Sheron et al. [20] proposed a technique for improving robotic object identification accuracy that they named projection-dependent input handling (PDIP). Despite their low error rate and excellent recognition ratio, there are substantial limitations due to the difficulty of PDIP implementation and possible difficulties in real-world application. Greater testing and validation of the method's reliability and adaptability are required in several settings. Agricultural crops can only be protected from pests and illnesses if researchers create more realistic, efficient, and long-lasting robotic systems based on the Internet of Things. Table 1 summarizes pertinent research possibilities.

3. Proposed Methodology

This device is groundbreaking in the agricultural field because AgriBotIQ solves all the problems farmers face today. AI and the ubiquitous Internet of Things allow AgriBotIQ to facilitate precision agriculture by providing real-time analysis, tracking, and management of crops. It can benefit many fields, such as sustainable agriculture, remote monitoring, pest identification, and data-driven decision-making. By helping farmers to reduce the use of toxic pesticides and improve the utilization of current resources, AgriBotIQ offers a significant effect in the battle against sustainable agriculture. Farmers can find important information about their land, crops, and surroundings with this instrument. This data will allow them to make better judgments and run their business more efficiently. And when it comes to AI, IoT, and robotics, AgriBotIQ is way out in front, paving the way for all sorts of exciting new developments in the farming industry. Contemporary farmers concerned about their influence on the environment, efficiency, and productivity should purchase an AgriBotIQ.

3.1 AgriBotIQ system architecture

The aforementioned method for crop protection is based on substantial study; the goal is to provide a persuasive and practical answer. The proposed layout uses state-of-the-art technology and various data-gathering techniques, as seen in Figure 1. The IoT is the starting point of the digital revolution for farming with respect to weather monitoring and irrigation. Soil temperature, moisture, and humidity are the three variables that these sensors measure in real-time. In addition, AI techniques are combined with visual signs and patterns to detect plant diseases early on, allowing for faster treatment. As a component of production evolution, studying the development and change of crops is crucial. The many components of the AI and robotics-based crop protection system work together to improve farming:

Data collection module: The system uses various sources to generate insights, including weather predictions, soil sensors, satellite imagery, and digital bug traps. With this much data, we can foretell when weeds will sprout, pests will attack, diseases will spread, and even the weather. This module aims to teach students how to create and understand machine learning models so that they can programmatically analyse data and get valuable insights. The technology deciphers the results of ML models and delivers actionable insights to aid in decision-making.

The decision-making module uses AI technologies to assess the collected data and select optimal crop protection strategies. Several aspects are taken into account when assessing the efficacy of pesticide applications. These include the crop type, its developmental stage, the severity of the insect, weed, and disease, and underlying ecological and risk factors. Furthermore, we consider possible increases in pesticide resistance and legislative limitations.

Autonomous robots with sensors and cameras can use this technology in their robotic modules. We can create the most efficient crop protection processes conceivable by teaching these robots to detect and eradicate pests, diseases, and weeds using AI algorithms. The monitoring module, which employs AI to make necessary adjustments in real-time, can track crop protection techniques' effectiveness. Maintaining this awareness guarantees peak performance and permits prompt actions whenever necessary.

The user interface module offers a user-friendly platform for stakeholders, including agricultural scientists, farmers, and others, to communicate with one another. Customers can enter data through this interface to tailor the results to their requirements. The system's recommendations and outcomes are displayed.AI-powered robotic crop protection devices provide more effective crop protection with less chemical input. The system provides data-driven suggestions for crop protection strategies to encourage sustainable agricultural practices and maximize crop yields.

If a graph G' and a set of traversals T create the same inspired traversal trind, the definitive option is the single T traversal that can reach G' trind with the allowed visited order of vertices.By limiting themselves to canonical possibilities, GPM algorithms eliminate wasteful computation by ensuring that no two exploration traversals end up at the same induced traversal. It is possible to transform canonical candidates into a distinct representation known as a canonical representative, or pattern, in this context. Canonical labeling is a common procedure in GPM methods for subgraph classification; it involves converting an induced traversal to its canonical representation.

4. Results and discussion

The dataset details the methods used to cultivate crops for every region, season, and kind of soil [21]. The data encompasses a wide range of crops, including cotton, onions, potato, wheat, rice, ragi, groundnuts, sugar cane, and banana, as well as pertinent agricultural and environmental aspects. Important characteristics covered by the dataset include cropping period (kharif, rabi, or entirety year), place of residence (state), cultivated area, type of soil (e.g., alluvial ground, black soil), chemicals application rates (high, low, an average), soil pH, ambient temperature (low, high, average), use of fertiliser levels (low, high, an average), price of crops, and amounts of rainfall (high, low, average).

4.1 Recognition ratio

The accuracy rate, or recognition ratio, measures how well the AgriBotIQ system distinguishes between healthy, diseased, or nutrient-deficient plants. It is the proportion of plants that the system properly identified relative to the overall number of plants that were evaluated. Mathematically, recognition ratio (RR) is referred to as the following equation (8),

In equation (8), the total number of plants that the AgriBotIQ system accurately classified as healthy, diseased, or nutrient-deficient is called 𝑡𝑃 (True Positives). The number of plants that the AgriBotIQ system mistakenly deemed healthy, diseased, or nutrient-deficient is called 𝑓𝑃 (False Positives). The AgriBotIQ system must maximize true

After calculating the recognition ratio, agricultural researchers and specialists can assess how the AgriBotIQ system predicts a user's nutritional and health status. A higher recognition ratio enhances the efficiency and lifespan of agricultural production (Fig. 3), proving the system's effectiveness and dependability in managing crops and protection.

4.2 Processing Time

The time indication illustrates the median length for the AgriBotIQ technology to analyze the data, find trends, and provide valuable insights. Very quickly and efficiently, the system completes the work measured by it. The method described in equation (9), which shows the sum of all execution times divided into the total of all observations, can be used to estimate the mean time for processing (AT) of the AgriBotIQ system:

Equation (9) shows that the total number of measurements is represented by 𝑡, and that each observation, indicated as𝐷𝑖 , requires iterations of the AgriBotIQ technology. Much work goes into each observation, including collecting data, cleaning it, identifying features, identifying trends, and producing insights. The data collected by the AgriBotIQ system may be examined via various methods such as artificial intelligence (AI), machine learning (ML), and others. To provide growers and agricultural professionals with timely insights, the AgriBotIQ platform may use efficient algorithms and appropriate system architecture to decrease the median processing time.

Reduced processing times improve cultivation and protection, as seen in Figure 4. This allows for quicker decision-making and reactions as crop conditions fluctuate. Scientists and engineers may be able to render AgriBotIQ smarter and more affordable if they monitor the median time measure closely. Improved farming methods that are friendlier to the environment will develop from that amount.

4.3 Error rate

An error rate provides a reliable indication of the number of times the AgriBotIQ system makes faults in crop analysis and monitoring. The manner in which the system works demonstrates whether it can distinguish between healthy, damaged, and nutritionally deficient crops. We may find the error rate (ER) by dividing the overall number of misidentifications by the total number of occurrences, as stated in Equation (10).

The acronym for the rate with which a specific state, like "healthy," "sick," or "deficient in nutrients," is correctly identified as "𝑡𝑃" in Equation (10). Can the state get the absence label correct some proportion of the time? True Negatives, abbreviated as𝑡𝑁, have that meaning. "Fake positives," or𝑓𝑃, is the frequency with which an incorrect diagnosis is produced when none exists. When a condition is mistakenly thought to be absent when it is present, this is called "False Negative," or𝑓𝑁. A false positive occurs when the AgriBotIQ technology mistakenly identifies abundant nutrients in healthy plants or excess nutrients in unhealthy plants. Such an occurrence occurs during the assessment and monitoring of crops. A false negative occurs when the system mistakenly labels a plant in good health as ill or lacking nutrients, even when the plant is perfectly OK.

According to the study, the AgriBotIQ system may be evaluated for accurate and wrong identifications by determining the mistake rate. Fig. 5 shows that a system with reduced error rates is more trustworthy and accurate regarding crop tracking and assessing activities. Researchers and developers may want to examine and tweak the error rate parameter to improve the AgriBotIQ system. Based on this data, better techniques for crop preservation and management will be accessible.