Re-forestation using “self-driven” robotic vehicle

Ms. VINUTHA V Ms. NANDINI R Ms . CHANDANA K

School of Computer Science and School of Computer Science and School of Computer Science and
Information Science Information Science Information Science
Presidency University Presidency University Presidency University
Bengaluru, India Bengaluru, India Bengaluru,India
VINUTHA.20211CIT0153@presidenc NANDINI.20211CIT0070@presidency CHANDANA.20211CIT0106@preside
yuniversity.in university.in ncyuniversity.in

Ms. REDDY SREYA Dr. Anandaraj S P Mr. E Sakthivel

School of Computer Science and School of Computer Science and line 2: School of Computer Science and
Information Science Information Science Information
Presidency University Presidency University Presidency University
Bengaluru,India Bnegaluru,India line 4: Bengaluru, India
REDDY.20211CIT0101@presidencyun [email protected] [email protected]
iversity.in

Abstract—Research on automated agriculture robotic Robotics and Machine Learning have evolved in different
vehicle is a crucial aspect of advancing the agriculture industry, industries, including agriculture. Both the technologies have
particularly in India, where agriculture serves as the backbone solved some of the major concerns such as labor crisis, crop
of the economy and significantly contributes to GDP. This paper failure, weather unpredictability, declining soil fertility,
introduces a novel approach to development of Re-forestation inefficient resource utilization, crop diseases and pest
using “self-driven” robotic vehicle integrated with software and infestation [3]. Majority of these are caused due to knowledge
hardware components. This primary aim of this research is to gaps and lack of proper technologies that can assist farmers
enhance the productivity and sustainability in reforestation [4].
practices by leveraging advanced technologies like a robotic
vehicle and Machine Learning Algorithms. Focused on Hence this research introduces an integrated system
technologies aspects of agriculture, this research paper presents combining a self-driven robotic vehicle and machine learning
an integrated system consisting of two main functionalities: algorithms to improve farming and reforestation practices.
1)Robotic vehicle that performs various farming activities that These technologies can address critical agricultural challenges
assist farmers such as ploughing, sowing , irrigation.2)Crop by automating labor-intensive tasks, improving decision
prediction using live sensors such as temperature , soil moisture making and increasing efficiency. The proposed system
and atmospheric pressure .The crops are predicted using the address crop yield prediction, adaptive crop recommendation
data sensed by these sensors and use Machine Learning
and large-scale afforestation using the self-driven robotic
Algorithms trained on historical field data.
vehicle [5].
Additionally this research work helps the beginner farmer Automated farming activities- The robotic vehicle
in sowing the adaptive crop reducing the man power, enhances automates farming tasks such as ploughing, sowing, irrigation
planting precision by automated planting system and reduces and weed removal , reducing the need of extensive labor and
the chances of crop failures through crop recommendation increasing operational precision [6].Data collected through
system. Finally the approach followed in this shows excellent sensors can be viewed in Blynk IoT [7]. Another feature of
potential for automating the farming practices and crop
the system is Suitable crop Prediction- By analyzing data
prediction. In fact, the test was performed on the rice and wheat
through sensors measuring temperature, soil moisture and
crops demonstrated the effectiveness of the proposed approach,
and the same approach can be used on other crops to achieve
atmospheric pressure, the system predicts crop suitability and
the same goal. recommend optimal crops to farmers [8] [9].
These technologies not only enhance planting precision
Looking ahead, this research offers insights into further but also help farmers adapt to changing environmental
developments of automating forest management, accurate yield conditions, mitigate crop failures, and increase yield. The
prediction of recommended crop, and maintaining the
system's applicability extends beyond conventional
ecological balance.
agriculture to reforestation efforts, enabling large-scale tree
Keywords—Reforestation, precision farming, crop prediction, planting and ecosystem restoration with minimal human
Machine Learning Algorithms, self-driven robot, farming intervention [10].
practices.

I. INTRODUCTION
Agriculture has been a cornerstone of human civilization,
to satisfy the basic daily needs. In India, agriculture serves as
primary occupation for a large sector of Indian population.
Several reasons such as changing weather conditions,
depletion of natural resources and pollution has significantly
reduced the crop production [1] .To address these concerns,
there is need for innovate technologies that can enhance
productivity and promote sustainable farming practices [2].

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

 II. LITERATURE REVIEW not address real-time or live data integration, which is crucial
for adaptive prediction systems. Additionally relaying only on
A. Songyu Li, Morgan Rossander, Hakan Lideskog the historical data may result in inaccurate prediction in
(June 2024) unforeseen natural conditions.
Proposed a system that performs obstacle detection and
providing estimated position to plant planner to function Sensor based monitoring or live monitoring of the
which in turn plan and select planting areas. Looking ahead, environmental conditions to predict the suitable crop and
future developments for irrigation, ploughing, weeding can be monitoring the crops in real time has not been possible in the
done in the same proposed methodology [6]. existing methods.

B. Michail Moraitis ,Konstantinos Vaiopoulos , Athanasios IV. PROBLEM STATEMENT

T.Balafoutis (February 2022) Reforestation efforts and traditional farming practices face
Developed an economic robotic system for the automatic challenges due to labor intensity, inefficiency, and
monitoring and management of urban gardens. The systems environmental degradation. Current traditional tasks such as
was integrated providing precise movement of its actuator and ploughing, sowing, irrigation, and weed removal are labor
applying precision irrigation based on specific needs. This intensive tasks, leading to inconsistent results. Additionally
system can be further more developed by deploying into crop failures caused by unpredictable environmental
farming and forest areas, by increasing its capabilities [4]. conditions hinder agricultural productivity and sustainability.
C. Abdellatif Moussaid ,Sanaa El Fkihi October 2022) Advancements in agriculture and reforestation robots have
made notable progress. However, there is a need for more
The main aim of the project was to define an intelligent
precise solutions. Current systems focus on limited
system for predicting citrus fruit yield before the harvest
functionalities such as obstacle detection and planting position
period using machine learning algorithm trained on historical
selection, absence of sensor based real time data monitoring
field data. Though this study shows excellent potential for
systems, adaption to rural farming and forest reforestation,
fruit yield prediction, the same method may not give accurate
inability to predict the crop suitable for the weather
results for rest of the crops [8].
conditions.
D. Tylek Pawek, Szewczyk Grzegorz (2023) To address these challenges, a need arises for a robotic
The paper presents design solution of the key working vehicle that can automate farming practices while ensuring
unit, which is universal, openable dibble, cooperating with a precision and efficiency controlled by RF module. The
three-toothed shaft to prepare a planting spot. The solution proposed method automates tasks from the initial phase of
enables continuous operation of the machine. This system can crop prediction-Utilizing live environmental data to
be future integrated with system for planting and other recommend the suitable crop, preparing the soil for optimal
farming activities such that fully functional system can be seed placement, planting seeds at precise intervals, and finally
deployed in a Forest [10]. monitoring the soil moisture and providing the appropriate
amount of water to crops.
E. Joy Iong-Zong Chen and Pisith Hengjinda (2019).
The purpose is to apply an artificial intelligence (AI) to This reforestation robot can operate through RF module in
manage the operation of a farm robot. The researchers have diverse terrain, adapt to varying soil and climatic conditions
designed robot for farming rice and its control system based and provide a sustainable farming solution for large-scale
on AI. The robot can move automatically with the data planting. It aims to increase the agricultural efficiency, reduce
collected. To complete the operation, power consumption is a labor dependency, and minimize environmental impact while
major concern [5]. contributing to forest reforestation.

III. DRAWBACKS OF EXISTING METHODS

Though there is advancement in technology in agriculture
and farming practices, there are drawbacks with respect to the
proposed method. Some of the systems focus only on the
obstacle detection and planting selection, neglecting other
essential farming activities like preparing the soil for optimal
seed placement, monitoring soil moisture and providing
appropriate amount of water to crops.
Other integrated systems relay on specific planting
mechanism (dibble with three-toothed shaft) may limit
adaptability to different soil types or terrains.
Some of the systems are limited to urban gardens, adoption
for forest or rural farming would require significant
modifications, which are not yet explored.
These systems also lack integration of advanced analysis
techniques for crop recommendations. Also these systems do
 prevents features with larger ranges from dominating those
V. METHODOLOGY with smaller ranges. Evaluation metrics to measure the
Reforestation robot address major agricultural issues by model’s performance are based on accuracy score and
automating farming practices and developing an application classification report (precision, recall, F1-score) .This
that can predict the suitable crop using Crop provides an understanding of how well the model is
Recommendation System. Following are the major performing on unseen data. For prediction, input features
developments: (e.g., soil properties, weather conditions, etc.) are first
scaled using the previously saved scaler.
1) Reforestation robotic vehicle – The hardware setup
integrates ESP32 NodeMCU as the central controller and For yield prediction, Random Forest Regressor with a
acts as coordinator between multiple processes. ESP32 Pipeline is used to predict the amount of yield. One-hot
NodeMCU is beneficial as it has dual-core processor which encoding, which coverts categorical variables to binary
can handle multiple tasks concurrently such as processing variables are used to process the data. Interactive prediction
the sensor data, controlling motors and managing and Pipeline usage are used for efficient model
communications. Additionally, it solves the problem of cost management.
and energy as they are cost and energy efficient.
VI. RESULTS
Servo motors are utilized for ploughing and seed sowing
as it has control of angular or linear position, and has The system effectively automates agricultural
compact size. Removal of unwanted plants which hinder the operations such as ploughing, sowing, irrigation, and
growth of the crops, this is done with the help of DC motor. weeding and ensures precise and consistent movements
enabling uniform field coverage. Real-time monitoring of
Sensors like soil moisture, DHT sensor for temperature live environmental data including soil moisture,
detection, and BMP sensor for pressure provide real-time temperature, and pressure allows optimum utilization of
environmental data. This data can be further utilized in Crop resources, with irrigation pump activity only when
Recommendation System. Soil moisture can be detected necessary, and minimizing water wastage. Real-time
and real time monitoring of the moisture levels can be done. feedback mechanism through Blynk IoT app enhances
When there is an indication saying the soil is dry, the system system transparency and user control. The robot also shows
has automated irrigation system which turns on. All the seamless field navigation. The system has enhanced the
sensor data are collected and stored in Blynk app. Sensor efficiency of agricultural tasks.
monitoring and control can also be done through the same
application. Crop recommendation and yield prediction can protect
the crops from crop failures through early detection. Crop
The system uses RF communication to control a robot recommendation suggests crops that can be best grown in
designed for field mobility. They work on the principle of the given agricultural conditions considering temperature,
transmitter and receiver. By combining automation and pressure, pH, Nitrogen, Phosphorus, humidity, rainfall.
remote mobility, the methodology reduces manual labor, Yield predictions are also done on machine learning
improves precision and enhances efficiency in agricultural algorithms.
operations, making it a robust solution for both rural, urban
farming and forest reforestation.

VII. DISCUSSION
The results of the system highlight effectiveness in
automating essential agricultural tasks, aligning with the
growing need for automated agricultural practises and
resource optimization. The use of NodeMCU ESP32, with
integration of RF module , not only ensured seamless
control and data monitoring but also demonstrated its cost-
2)Crop and Yield Recommendation system: The system
effectiveness and scalability compared to traditional
uses Random forest classifier to build a machine learning
microcontrollers. The precise operation of servo motors in
model for crop recommendation. Feature values are
ploughing and seed sowing underscores their suitability for
standardized using Standard Scaler, which ensures that the
agricultural automation, while real-time data from soil
data is centered and scaled for optimal performance. This
moisture and environmental sensors enabled smart
irrigation, reducing water usage and promoting XI. REFERENCES
sustainability. The RF-controlled robotic mobility further
emphasized the potential for human intervention in
inaccessible or hazardous areas, ensuring flexibility and [1] F. a. A. O. (FAO), "The State of Food and
safety.These findings corroborate existing knowledge on Agriculture 2020: Overcoming Water Challenges in
the benefits of automation and IoT in agriculture, while also Agriculture".
showcasing how modular, energy-efficient systems can [2] P. S. a. N. G. R. Singh, "Emerging Technologies in
improve productivity and reduce labor dependency. The Agriculture," Journal of Agricultural Sciences, vol.
system’s adaptability to diverse field conditions and its 27, p. 5, 2019.
potential integration with Blynk IoT app suggest broader
implications for scalable smart farming solutions, bridging [3] J. W. a. Q. L. Y. Zhang, "Application of Robotics in
the gap between traditional practices and modern Agriculture: Challenges and Opportunities,"
technological advancements in agriculture [11]. International Robotics Journal, vol. 18, pp. 105-
121, 2021.
Random forest classifier is a supervised machine
[4] K. V. a. A. T. B. M. Moraitis, "Design and
learning algorithm used for classification tasks. It is based
on the concept of ensemble learning, where multiple Implementation of an Urban Farming Robot,"
decision trees are trained on random subsets of the data, and Agricultural Robotics Quarterly, vol. 5, pp. 45-58,
their outputs are aggregated to make the final prediction. 2022.
Hence it is best suited for crop prediction system for [5] J. I. Z. C. a. P. Hengjinda, "Applying AI
accurate results [12]. The Random Forest Regressor is a Technology to the Operation of Smart Farm Robot,"
supervised machine learning algorithm used for regression Artificial Intelligence in Agriculture, vol. 4, pp. 12-
tasks. Like the Random Forest Classifier, it also uses an 25, 2019.
ensemble of decision trees hence, best suited for yield
[6] S. L. a. H. L. M. Rossander, "Vision-Based Planting
prediction [13].
Position Selection System for an Unmanned
Reforestation Robot," Forestry Robotics Journal,
vol. 12, pp. 33-45, 2024.
VIII. CONCLUSION
[7] M. S. a. P. Gupta, "Automation and IoT: Bridging
The system effectively automated ploughing, seed technology and agriculture," Computers and
sowing, and irrigation, enhancing precision and reducing Electronics in Agriculture, pp. 102-112, 2020. .
manual labor. Using NodeMCU ESP32, sensors, and servo
motors, it demonstrated cost-effective and energy-efficient [8] A. M. a. S. E. Fkihi, "Machine Learning Applied to
smart farming, with real-time environmental monitoring Tree Crop Yield Prediction Using Field Data and
optimizing resource utilization. The RF-controlled robot Satellite Imagery: A Case Study in a Citrus
added flexibility for field navigation, making the system Orchard," Agricultural Informatics Review, vol. 8,
adaptable and scalable. The system integrated with crop and pp. 89-102, 2022.
yield recommendation system can suggest the suitable crop [9] A. R. J. a. D. S. Patil, "An Approach Towards Crop
and same from crop failures. Recommendation System Using Random Forest
Machine Learning Algorithm - A Review".
[10] P. T. a. G. Szewczyk, "Design of a Planting Module
IX. FUTURE ENHANCEMENT for an Automatic Device for Forest Regeneration,"
For future improvements, autonomous navigation using Journal of Forest Engineering and Automation, vol.
GPS or machine vision and cloud-based data analytics can 15, pp. 77-88, 2023.
be incorporated for larger-scale applications. Adding
sensors like pH and nutrient monitors could enable
comprehensive soil analysis. This system provides a robust
foundation for advancing precision agriculture and
sustainable farming technologies. Crop recommendation
and yield prediction system can be improved by
incorporating more features and predicting more accurate
results.
X. ACKNOWLEDGMENT
We would like to sincerely thank our guide, Dr.
Anandaraj S P, support and guidance throughout the project.
We also extend our gratitude to our institution Presidency
University for providing the resources and assistance we
needed, on time. Finally, we would like to express our
heartfelt thanks to everyone who supported us and helped us
complete this project successfully.