HORTICBOT - AN INTEGRATED CROP HEALTH AND PEST MANAGEMENT WITH NDRE ANALYSIS

Abstract

HORTICBOT An Integrated Crop Health and Pest Management Bot with NDRE Analysis ABSTRACT: The present invention pertains to the field of agricultural technology and introduces an autonomous agricultural robot known as the Horticbot. The Horticbot is an innovative, multifunctional agricultural solution designed to address the challenges of crop monitoring, pest and .weed detection, and precise organic pesticide application, while also facilitating comprehensive data collection for crop growth analysis. Leveraging a combination of advanced sensors, image processing, and artificial intelligence algorithms, the Horticbot locomotes through agricultural fields to capture, process, and analyze images of crops, thereby enabling real-time identification of pests and weeds. Additionally, it measures critical growth parameters, including. plant height, stem width, chlorophyll content, and leaf density, at periodic intervals to determine the growth of plants. The Horticbot integrates advanced technology for precise crop management, optimizing productivity, resource efficiency, and eco-friendly practices.

Specification

FORM 2 THE PATENTS ACT, 1970
(39 of 1970)
&
The Patents Rules, 2003
PROVISIONAL/COMPLETE SPECIFICATION
(See section 10 and rule 13)

1. TITLE OF THE INVENTION: HORTICBOT - An Integrated Crop Health and Pest Management with NDRE Analysis

2. APPLICANT(S)
APPLICANTS NAME
NATIONALITY
ADDRESS
1. Sri Sairam Engineering College
Indian
ALL HAVING ADDRESS AT
2. Vishwa M
Indian
Sri Sairam Engineering College,
3. Swetha A
Indian
Sai Leo Nagar,
4. Hari Haran B
Indian
West Tambaram,
5. Ms. R Chitra
Indian
Chennai-600044

3. PREAMBLE TO THE DESCRIPTION
CD □) ra Q.
CD
PROVISIONAL
The followrng-specifrcation-dcscribes the invention:
COMPLETE
The following specification particularly describes the invention and the manner in which is to be performed.
4. DESCRIPTION (Description shall start from next stage.)
Enclosed - Along with this form
cxi
E
o
5. CLAIMS (Not applicable for provisional specification. Claims should start with the preamble
- "I/We claim" on separate page)
Enclosed - Along with this form
6. DATE AND SIGNATURE (to be given at the end of the last page of the specification)
O ID
O CO co o
Tf Tf
CX| o CXJ co
Tf CXI co
7. ABSTRACT OF THE INVENTION (to be given along with complete specifications on separate page)
Enclosed - Along with this form

Note:-
*Repeat boxes in case of more than one entry.
*To be signed by the applicant(s) or by an authorized registered patent agent.
*Name of the applicant should be given in full, family name in the beginning.
*Complete address of the applicant should be given stating the postal index no./code, state and country.
*Strike out the column(s) which is/are not applicable.

HORTICBOT
An Integrated Crop Health and Pest Management with NDRE Analysis

FIELD OF INVENTION:
The field of invention for Horticbot resides at the intersection of several key domains, synergistically orchestrated to create a revolutionary precision agriculture and advanced crop care automation system. These domains encompass the following:

1. Image Processing:
Domain Utilization To analyze the data obtained by multispectral sensors, Horticbot employs image processing techniques.
Application: Image processing techniques are employed to enhance data quality, extract useful information, and produce NDRE (Normalised Difference Red Edge) maps for assessing crop health and vitality.

2. Multispectral Sensors:
Domain Utilization: Red and near-infrared (NIR) bands are just two of the wavelengths that multispectral sensors that may collect data.
Application: Horticbot accurately determines the amount of chlorophyll in plant leaves by using multispectral data for NDRE calculations. The system's data-collecting mechanism is based on these sensors.

3. NDRE Calculation:
Domain Utilization: The computations behind NDRE (Normalised Difference Red Edge) are at the core of Horticbot's technology.
Application: The chlorophyll concentration of plant leaves, which is a clear sign of the health of a crop, is calculated using NDRE.

4. Robotics and Bot Design:
Domain Utilization: Horticbot encompasses the field of robotics, incorporating the design and construction of an agricultural automation system.
Application: The system includes a precisely built bot with robotic arms for weed removal, pest identification and precision farming. The bot's design guarantees focused and effective functioning in agricultural fields.

5. Artificial Intelligence (Al) and Image Recognition:
Domain Utilization: The system is employed with Al and image recognition technology for pest detection.
Application: Advanced image recognition algorithms are used by Horticbot's pest detection model, leaf disease model, and NDRE Value Detection model to identify and classify common pests and diseases impacting crops. The Al-powered model helps with accurate intervention and early diagnosis.

Horticbot encompasses an array of domains, each contributing to the creation of an advanced agricultural automation system with a holistic approach to crop care, pest detection, and weed management. The integration of these domains results in a versatile and innovative technology poised to revolutionize agriculture for the better.

BACKGROUND OF INVENTION:
Harshita Nagar et al. (2021) introduced a method for automatic pest detection utilizing Wavelet transformation and Oriented FAST and Rotated BRIEF (ORB). The proposed approach is demonstrated on images of fluffy caterpillar pests on mustard crop and fava bean crop farms in Rajasthan. The Region of Interest is extracted using wavelet transformation and image fusion techniques.

Harshita Nagar and R.S. Sharma et al. (2020) conducted a comprehensive survey on pest detection techniques using Image Processing which presents the various image processing techniques such as feature extraction and automatic detection for the image. The survey shows the efficient and simple existing methodologies. Several techniques were used to obtain knowledge of different background modeling for pest detection such as image filtering, median filtering for noise removal, image extraction and detection through scanning. It also depicts some promising results to present enhanced methods and tools for creating fully automated pest identification including extraction with detection.

M. I. Pavel, et al (2019) implemented Image processing techniques to detect and classify the affected plant disease. In the process, the work is divided into four portions which are image acquisition, preprocessing, segmentation of affected regions, feature extraction, and classification using a multi-class support vector machine algorithm.

Califano, F.; Cosenza, C.; Niola, V.; Savino, S. (2022) et al. The multibody rover structure is designed by employing a sequence of rigid bodies, flexible bodies, and joints. Some joints are connected to motors that drive the motion of the rover parts and actuation system. In this way, in the model, there is a connection between the rotational mechanics and the electrical domain. There is a multibody model of the ground. The ground is composed of several rigid bodies.

SUMMARY:
The invention, an autonomous agricultural robot referred to as the Horticbot, is designed to transform crop monitoring and management in agriculture. The Horticbot incorporates advanced sensors, image processing, and artificial intelligence to autonomously capture and analyze crop images, facilitating real-time identification of pests and weeds. Furthermore, it measures key growth parameters, including plant height, stem width, chlorophyll content, and leaf density. This invention offers a holistic approach to precision agriculture, boosting crop productivity while reducing resource usage. It combines robotics, image analysis, and Al to provide.an integrated, efficient, and environmentally conscious solution for modem farming practices.

OBJECTIVES OF PROPOSED INVENTION:
The primary objective of this invention is to provide an autonomous agricultural robot, the Horticbot, which addresses the following key goals:

1. Precision Crop Monitoring: To autonomously capture and analyze images of crops, enabling real-time identification of pests and weeds, as well as measurement of essential growth parameters (plant height, stem width, chlorophyll amount) with a high degree of precision.
2. Efficient Pest Management: To autonomously apply pesticides only in response to detected pest-infested areas, thereby reducing the indiscriminate use of chemicals and promoting eco-friendly farming practices.
3. Comprehensive Data Collection: To facilitate the collection, storage, and analysis of extensive data related to crop growth, supporting data-driven decision-making for crop management.
4. Enhanced Crop Productivity: To streamline crop monitoring and management processes, ultimately enhancing crop productivity, optimizing resource utilization, and promoting sustainable agricultural practices.

By achieving these objectives, the Horticbot aims to significantly improve crop management practices and contribute to sustainable, environmentally responsible farming.

DETAILED DESCRIPTION OF THE INVENTION:
Introduction: The present invention, referred to as "Horticbot," is a groundbreaking agricultural automation system designed to address the challenges associated with precision crop care, pest detection, and weed management. Horticbot is ideally suited for the cultivation of key crops such as brinjal (eggplant), green chillies and tomatoes and it integrates a range of innovative technologies to revolutionize modem farming practices.

Choice, of Target Crops: The selection of brinjal, green chillies, and tomatoes as the primary focus of Horticbot is driven by various factors:

• Economic Significance: These crops are important agricultural commodities in many foods and have a substantial economic impact both domestically and globally.
• Diverse Pest and Disease Profiles: Each of these crops is susceptible to a range of pests and diseases, making them excellent candidates for showcasing Horticbot's pest detection capabilities.
• Growth Characteristics: Brinjal, green chillies, and tomatoes exhibit diverse growth patterns, resource requirements, and optimal environmental conditions. Horticbot can be adapted to accommodate these variations, making it versatile for a range of crop types.

Innovative Technology: Horticbot leverages cutting-edge technology to accomplish its objectives. Key technological components include:
• Convolutional Neural Networks (CNNs): These neural networks are the backbone of Horticbot's image-processing capabilities. CNNs are adept at feature extraction and pattern recognition, making them ideal for tasks such as identifying pests and diseases in crops.
• Multispectral Sensors: The multispectral sensors integrated into Horticbot collect data in specific spectral bands, including the red and. NIR regions. The collected data is fundamental for calculating NDRE values, which serve as a direct indicator of chlorophyll content and, consequently, crop health.
• Autodesk Fusion 360: The use of Autodesk Fusion 360 for designing the bot's structure and components ensures precise engineering, robust construction, and efficient functionality.
• Real-time Data Processing: Horticbot processes data in real time, enabling instant assessments of crop health and the prompt detection of pests and diseases. The system harnesses advanced algorithms to perform NDRE calculations and image recognition.

Bot Design and Construction: The bot's design is an integral component of Horticbot's success. Detailed considerations are made to accommodate the following aspects:
• Robotic Arms: Horticbot is equipped with robotic arms designed to handle tasks such as weed removal. These arms are meticulously engineered to ensure optimal performance and minimal interference with crop growth.
• Sensors integration: The placement and integration of multispectral sensors are designed to minimize interference with crop canopies while ensuring accurate data collection.
• Stability and Mobility: Horticbot is engineered for stability and mobility in various field conditions. Its design includes features such as adjustable heights, robust wheels or tracks, and adaptability to terrains.

Data-Driven Precision Agriculture: Horticbot empowers precision agriculture through data-driven decision-making:
• NDRE Calculations: Horticbot calculates NDRE values based on multispectral data, providing an instant assessment of chlorophyll content and crop health.
• Pest Detection and Weed Removal: Utilizing image recognition and CNNs, the system identifies and classifies pests, diseases, and weeds. Early detection allows for timely intervention.
• Resource Optimization: Data-driven recommendations are provided for irrigation, fertilization, and pesticide application. This approach optimizes resource utilization, leading to increased crop yields and reduced environmental impact.

Sustainable Agriculture:
The implementation of Horticbot promotes sustainable agriculture practices:
• Reduced Chemical Usage: By enabling precise, targeted intervention, Horticbot reduces the need for chemical treatments and minimizes their impact on the environment.
• Efficient Resource Allocation: Resource optimization through data-driven recommendations contributes to the efficient use of water, nutrients, and pesticides.
• Enhanced Crop Health: Horticbot's capabilities, in early pest detection and weed management help maintain crop health, leading to improved yields and sustainability.

NDRE IMPLEMENTATION:
The implementation of NDRE (Normalized Difference Red-Edge) in Horticbot plays a pivotal role in assessing the chlorophyll content and overall health of crops. The report provides a detailed account of how NDRE is calculated and integrated into the Horticbot system for real-time precision agriculture. The NDRE formula is a modified version of the normalized difference index, specifically tailored to capture the red-edge region of the electromagnetic spectrum. The formula is defined as follows.

NDRE = (NIR-RE)/(NIR+RE+£)

Where
NIR: Near-infrared reflectance values from the crop.
RE: Red-edge reflectance values from the crop.
e: A small constant (l*10A-8) to prevent division by zero.

Multispectral Sensor Data Acquisition: Horticbot is equipped with multispectral sensors capturing data in specific spectral bands, including the red and NIR regions.

Data Preprocessing: The obtained multispectral data is preprocessed to extract the NIR and RE bands, which are crucial for NDRE calculation.

NDRE Calculation Algorithm: Horticbot's algorithm implements the NDRE formula, computing NDRE values for each pixel in the captured images.

Integration with Real-time Data Processing: NDRE calculations are seamlessly integrated into Horticbot's real-time data processing capabilities, enabling instant assessments of crop health.

Utilizing Different Shades for Enhanced NDRE Calculation: In addition to the standard NDRE calculation, Horticbot implements a unique approach involving the utilization of different shades of green in the images. This enhancement aims to:

Capture Varied Chlorophyll Content: Different shades of green represent variations in chlorophyll content. This allows Horticbot to differentiate between leaves with high and low chlorophyll levels.

Enhance Precision in Health Assessment: By considering a spectrum of green shades, Horticbot achieves a more nuanced NDRE calculation, leading to a more precise assessment of crop health.

Conclusion: Horticbot is a pioneering invention that combines cutting-edge technology, precision design, and data-driven agriculture to address the complex challenges of modem crop management. It has the potential to transform farming practices and promote sustainability in agriculture, especially in commercially significant crops, by effectively identifying pests, managing weeds, and boosting crop vitality.

BRIEF DESCRIPTION OF DRAWINGS:
Fig 1: Block diagram of Horticbot User Interface: This block represents the interface through which users interact with the system. It could be a graphical user interface (GUI) on a computer, a mobile app, or any other means through which users can control and monitor the Horticbot.

Software Module: This block encompasses the software components responsible for coordinating the various functionalities of the Horticbot. It may include the main control algorithm, communication protocols, and the overall logic that governs the robot's actions.

Image Capture: This module involves capturing images using cameras mounted on the Horticbot. These images are likely used for various purposes such as pest detection, leaf disease identification, and crop health assessment.

Image Processing: The captured images are processed, using algorithms and techniques tailored for agricultural applications. Image processing aims to enhance the quality of images, identify objects of interest, and extract relevant information for analysis.

Data Analysis: This block involves the analysis of the processed data, which may include information about pest presence, disease severity, and overall crop health. Data analysis algorithms interpret the images and provide valuable insights for decision-making.

Reporting: The results from data analysis are presented in a user-friendly format, possibly through reports or visualizations. This information helps users make informed decisions about crop management strategies. Pest Detection, Leaf Disease Detection, and Crop

Health Assessment: These blocks represent specific functionalities of the system. Each involves algorithms and processes tailored to detect and assess .the health of crops, identify pests, and diagnose leaf diseases.

Weed Detection and Growth Monitoring: Similar to the previous blocks, this section focuses on algorithms and processes designed to identify and monitor the growth of weeds in the agricultural field.

Communication: This module facilitates communication-between the Horticbot and external systems. It could involve data transmission, remote control, or reporting back to a central server or user interface.

Mobility, and Navigation-. This block represents the components responsible for the movement of the Horticbot. It may include motors, wheels, or tracks for mobility, as well as navigation systems like GPS for location tracking.

Bot Design (Physical Structure): This block includes the physical components of the robot, such as the chassis, frame, and any other structural elements that make up the Horticbot.

Sensors (GPS, Proximity): These are the sensors integrated into the Horticbot for gathering environmental data. GPS helps in determining the robot's location, while proximity sensors may be used for obstacle detection.

Power Supply (Rechargeable Battery): This block represents the power source for the Horticbot, typically a rechargeable battery that provides the necessary energy for the robot to operate.

Fig.2: Accuracy graph of Pest Detection

The accuracy of the pest detection is 96.44% which indicates the effectiveness and accuracy of Horticbot to correctly detect the presence of pests and classify them based on actual instances of pests in the test datasets. The X-axis of the graph represents the independent variable, which is the number of iterations or training cycles of the algorithm over a period and the Y-axis represents the dependent variable, which is the accuracy of the pest detection algorithm. Initially, the algorithm's accuracy was quite low, measuring below 20%. This low accuracy might be due to several reasons, such as inadequate training data, ineffective feature extraction, or an insufficiently trained model. To improve the accuracy of the algorithm, multiple iterations of training and refinement were performed. As the algorithm underwent several training cycles and refined its internal representation of pest detection patterns, its accuracy steadily improved. This gradual increase in accuracy is likely reflected in the graph as an upward trend. The algorithm managed to achieve an impressive accuracy rate of 96.44%.

Fig 3: Accuracy graph of Leaf disease detection
The accuracy of the leaf disease detection is 80.67% which represents the ability of Horticbot to precisely identify specific leaf diseases. The Y-axis typically shows accuracy values ranging from 0% (completely inaccurate) to 100% (perfect accuracy). The values in the upward trend indicate improved accuracy while the values in the lower trend represent lower accuracy. At the outset of the algorithm's development, its accuracy was below 30%. This low accuracy is due to insufficient training data and inefficient feature extraction. As the algorithm underwent multiple iterations, the accuracy gradually started to increase contributing to an upward trend. After a series of iterations, the algorithm achieved an accuracy rate of 80.67%. This signifies that the algorithm has become proficient at detecting tomato leaf diseases, with the majority of test cases correctly classified. This graph reflects the algorithm's potential for practical use in the field of plant disease management, with further opportunities for optimization and enhancement.

Fig 4: Chilli cultivation

The figure depicts a chilli cultivation method organized in a row fanning configuration, where chili plants are arranged in rows with a maximum spacing of 2 feet between them. This specific layout allows for systematic and efficient chilli cultivation, promoting optimal utilization of available space and resources. The organized rows facilitate ease of access for monitoring and management, ensuring precise crop care and yield optimization. The 2-foot spacing provides adequate room for Horticbot to locomote through it.

Fig 5. Design of Horticbot & Fig 6. Top view of Horticbot
The figures provided showcase the design of the Horticbot, with specific measurements and component features detailed as follows:
• The dimensions of the body of the Horticbot are 60 cm in length and 30 cm in breadth, with a height of 17.5 cm.
• The body structure comprises two separate containers, each designed to accommodate organic pesticides. Each of these pesticide containers is rectangular and measures 10 cm in width, 10 cm in depth, and 20 cm in height, with a capacity of 2 liters for pesticide storage.
• The front section of the Horticbot design is equipped with a rod designed to support the placement of a camera responsible for capturing images of the crops.
• In the rear section of the design, two rods are incorporated to facilitate the uniform and controlled spraying of pesticides.

Fig 7: Pest Detection - Bollworm
Pest Type: Bollworm
Detection Output:
This image illustrates the results of the pest detection model designed to identify bollworm infestations in crops.
Significance:
Bollworms cause damage to the reproductive structures of plants, and early detection aids in implementing effective pest management strategies.

Fig 8: Pest Detection - Stem borer
Pest Type: Stem Borer
Detection Output:
The image showcases the detection output of the pest detection model specifically trained for identifying stem borers in crops. Areas highlighted 'or marked in the image represent the presence of stem borers on the plant.
Significance:
Early detection of stem borers is crucial to prevent damage to the crop and allows for timely intervention, ensuring healthier plant growth.

Fig 9: Leaf Disease Detection - Tomato Mosaic Virus
Disease Type: Tomato Mosaic Virus
Detection Output:
The image represents the outcome of the leaf disease detection model specialized in identifying the Tomato Mosaic Virus in tomato plants.
Highlighted areas indicate the presence of the virus on the tomato leaves.
Significance:
Early detection of diseases like the Tomato Mosaic Virus is crucial for preventing its spread and minimizing yield losses in tomato crops.

Fig 10: Leaf Disease Detection - Tomato Early Blight
Disease Type: Tomato Early Blight
Detection Output:
This image showcases the results of the leaf disease detection model tailored to identify Tomato Early Blight in tomato plants. Marked regions in the image indicate the presence of early blight symptoms on the tomato leaves.
Significance:
Early detection of diseases like Tomato Early Blight supports timely and targeted management practices, contributing to overall crop health.

Fig 11&12 The Near-Infrared (NIR) spectrum illustrates the reflectance values of the crop in the near-infrared region. High values in the NIR spectrum suggest strong reflectance of near-infrared light, which is indicative of healthy vegetation. Regions with vibrant green shades represent areas where the crop reflects significant NIR light.

The Red-Edge (RE) spectrum displays the reflectance values of the crop in the red-edge region 'of the electromagnetic spectrum. Elevated values in the RE spectrum signify increased reflectance in the red-edge region, often associated with chlorophyll absorption. The spectrum reveals variations in shades of red, offering insights into chlorophyll distribution across the crop.

Fig 13 & 14 NDRE with Green Shades:
04jNov-2024/132446/202441083950/Form 2(Title Page)
The NDRE image incorporates different shades of green to represent varying chlorophyll content in the crop. Darker green areas indicate regions with lower chlorophyll content, possibly suggesting stress or nutrient deficiency. Brighter green'and yellowish areas correspond to healthier vegetation with higher chlorophyll levels. The accompanying NDRE values provide quantitative information on the chlorophyll status, aiding in precise crop health assessment.

Interpretation and Insights:
Correlation Between NIR and Healthy Vegetation: The strong reflectance in the NIR spectrum aligns with healthy vegetation, confirming the vitality of the crop.
Chlorophyll Distribution in RE Spectrum: The RE spectrum reveals variations in chlorophyll distribution, with distinct shades of red indicating regions of chlorophyll absorption.

NDRE Shades and Chlorophyll Content: Darker NDRE shades correspond to areas with potential stress or lower chlorophyll content. Brighter NDRE shades signify healthier vegetation with optimal chlorophyll levels.
Quantitative NDRE Values: NDRE values provide a quantitative measure of chlorophyll content, facilitating data-driven decision-making for targeted interventions.

CLAIMS:
We Claim,

Claim 1: A precision crop care and management system comprising
• a multispectral sensor for collecting data in the red and near-infrared (NIR) bands;
• an image processing module for analyzing multispectral data and calculating NDRE (Normalized Difference Red Edge) values to assess crop health.

Claim 2: The precision crop care and management system of Claim 1, further comprises a pest detection model that employs advanced image recognition techniques to identify and classify pests and diseases in crops.

Claim 3: The precision crop care and management system of Claim 1, is equipped with a weed removal mechanism comprising robotic arms and tools for detecting and removing weeds.

Claim 4: The precision crop care and management system of Claim 1, integrated with a precision agricultural module provides data-driven recommendations for irrigation, fertilization, and pesticide application based on NDRE data.

Claim 5: The precision crop care and management system of Claim 1, wherein the pest detection model achieves an accuracy rate of 96.44% in pest and disease identification.

Claim 6: The precision crop care and management system of Claim 1, further comprising a leaf disease identification model achieves an accuracy rate of 80.67% in identifying leaf diseases.

Claim 7: A precision crop care and management method comprising
• collecting multispectral data using a multispectral sensor;
• processing the multispectral data to calculate NDRE values for crop health assessment.

Claim 8: The precision crop care and management method of Claim 7, further comprising employing an advanced image recognition model for detecting and classifying pests and diseases in crops.

Claim 9: The precision crop care and management method of Claim 7, incorporates a weed removal mechanism with robotic arms and tools for detecting and removing weeds.

Claim 10: The precision crop care and management method of Claim 7, provides data-driven
recommendations for precision agriculture, including irrigation, fertilization, and pesticide
application, based on the calculated NDRE values.

Claim 11: The method of claim 7, wherein the method is. applied to the cultivation of specific
crops, including brinjal, green chillies, and tomatoes


Date : 30 .2m
Signature:
)\ Principal 2)
SAIRAM ENGINEERING COLLEGE
West Tambaram, Chennai-44.
1. Dr. J Raja
2. Vishwa M
3. SwethaA
4. Hari Haran B
5. Ms. R Chitra

Documents