Development of a Multipurpose Agricultural Robotic System: An Experimental Study

Muthukaruppan Vinaitheerthan¹ , Shri Dhanush Arul Mythili¹, Mirnalini Selvarajan Sivakumar¹ and Shruthi Senthil Raj²
1. Department of Mechanical Engineering, Amrita School of Engineering, Coimbatore, Tamil Nadu, India
2. Department of Agriculture Engineering, Kalaignar Karunanidhi Institute of Technology, Coimbatore, Tamil Nadu, India
Correspondence to: Muthukaruppan Vinaitheerthan,  vmuthucontact@gmail.com

DOI: https://doi.org/10.70389/PJS.100205

Additional information

• Ethical approval: N/a
• Consent: N/a
• Funding: No industry funding
• Conflicts of interest: N/a
• Author contribution: Muthukaruppan Vinaitheerthan, Shri Dhanush Arul Mythili, Mirnalini Selvarajan Sivakumar and Shruthi Senthil Raj – Conceptualization, Writing – original draft, review and editing
• Guarantor: Muthukaruppan Vinaitheerthan
• Provenance and peer-review: Unsolicited and externally peer-reviewed
• Data availability statement: N/a

Keywords: Multipurpose agricultural robot, Raspberry Pi-based farm automation, Sensor-fusion, Soil-monitoring, Machine-vision weed control, Low-compaction field robotics.

Peer Review
Received: 14 August 2025
Last revised: 22 October 2025
Accepted: 17 December 2025
Version accepted: 3
Published: 8 January 2026

Plain Language Summary Infographic

Introduction

Agriculture is a fundamental sector of human civilization, playing a crucial role in a nation’s economy. It significantly contributes to the supply of food, fiber, and other necessities for human consumption.¹ Farmers and agricultural professionals are slowly automating their farming methods to increase the productivity of their operations.² Technological advancements have enabled farmers to perform their operations and manage crops and livestock more effectively. By utilizing this technology, the farmers can track the crop growth details accurately. The new automated technologies have enabled farmers to maximize their resource efficiency and minimize operational costs while having a limited impact on the environment.³

This paper primarily explores the benefits of deploying Multipurpose agricultural robots in agriculture. Unlike specialized machines designed for single tasks, a multifunctional platform can integrate operations such as sowing, weeding, harvesting, and soil monitoring cost-effectively and sustainably. It examines how these robots can assist in achieving higher crop yields, reducing labor costs and soil compaction, and increasing farm operation efficiency.^4,5 It also discusses the challenges of implementing this technology and its potential impact on the future of agriculture.

Literature Review

Early developments in agricultural robotics focused on applying automation to reduce labor-intensive tasks while improving precision. Shamshiri et al. highlighted advances in crop scouting, autonomous harvesting, and weed control, emphasizing the role of integrated sensors, AI, and swarm-based systems in transforming digital farming. Their work showed how Unmanned Aerial Vehicles and Unmanned Ground Vehicles equipped with computer vision and machine learning could detect pests and enable site-specific pesticide spraying, setting the stage for precision agriculture.⁶ Building on automation in crop establishment, Naik et al. proposed a precision agriculture robot specifically designed for automated seed dispensing. Their prototype employed an algorithm to regulate seed placement and demonstrated accurate seeding performance under greenhouse conditions. This work illustrated how data-driven management of sowing can reduce input waste and improve yield consistency, representing one of the early practical applications of robotics in controlled agricultural environments.⁷

Parallel research also explored environmental monitoring for agriculture. Hwang et al. investigated wireless sensor networks for real-time data collection of temperature, humidity, and light levels in farm settings. Their study concluded that IoT-enabled monitoring systems were both cost-effective and suitable for continuous observation of field conditions, providing timely information for farmer decision-making. This work laid the foundation for integrating sensing networks with agricultural robotics, highlighting the synergy between environmental monitoring and autonomous decision support.⁸ Recent advances in agricultural robotics have emphasized lightweight, modular platforms designed specifically for smallholder farms. Yamasaki et al. developed an autonomous sensing system with high-accuracy crop scouting capabilities, highlighting the importance of compact and energy-efficient robots for field monitoring.⁹ Jin et al. and You et al. demonstrated how deep learning and image segmentation techniques have improved weed identification accuracy beyond 90%, a crucial step for autonomous weeding.¹⁰

Commercial robots such as Naïo Oz, FarmDroid FD20, and TerraSentia have already shown the viability of autonomous platforms in agriculture. Naïo Oz is a lightweight robot primarily focused on mechanical weeding and navigation between rows, while FarmDroid FD20 automates sowing and weeding using GPS-based precision but remains costly and targeted at large-acre farms. TerraSentia, designed for crop phenotyping, specializes in collecting plant trait data rather than direct field operations. In contrast, the proposed Multipurpose Agricultural Robotic System (MARS) is designed as a multi-functional, low-cost solution, integrating sowing, weeding, harvesting, and soil monitoring into a single lightweight platform. Unlike these commercial systems, MARS is optimized for small-scale Indian farms (<1 acre), where heavy machinery is impractical and unaffordable, thereby filling a critical gap between high-end agricultural robots and traditional manual labor.

Methodology

Soil Testing

Soil testing is the process of examining the soil condition. A DHT11 Temperature and Humidity sensor is used to measure the air humidity and temperature levels.¹¹ Additionally, an EZO pH sensor is attached to measure the pH level of the soil. These sensors are interfaced with Raspberry Pi 4B using GPIO pins with the help of an Analog-to-Digital Converter (ADC) to convert analog signals from the sensor into digital data for the Raspberry Pi to interpret. The Raspberry Pi 4B is programmed for the real-time analysis of the DHT11’s incoming data, which allows for the quick tuning of other automated processes.¹²

Farmers determine soil conditions by monitoring these variables, and based on the data provided by the sensors, the type of crop that should be grown can be identified. This is where the processing power of the Raspberry Pi comes in, making complex decisions on the sensor data for optimal resource utilization and more sustainable agricultural methods.^13,14 Validation trials were carried out on a 5 × 5 m demonstration plot planted with maize. Each task – sowing, weeding, and harvesting was repeated five times under loamy soil conditions (pH 6.7–6.9, soil moisture 18–22%) at an ambient temperature of 28–32 °C and 55–65% relative humidity. Performance metrics such as sowing spacing error, germination rate, weed detection accuracy, and harvesting damage rate were recorded and reported as mean ± SD across trials.

Sowing

During the sowing process, the robot drops the seeds one by one. The servo motor provides exact angular position control. By scattering seeds one at a time, farmers can control the seeds planted per unit area, which can improve crop yield and reduce seed consumption through the crank-slider lever mechanism.¹⁵ The servo motor MG996R is a high-torque motor used for precise angular position control for seed dispensing.

The Raspberry Pi 4B controls the servo motor using pulse-width modulation, a technique commonly employed by microcontrollers like the Raspberry Pi to control motors by varying the duty cycle of electrical signals. A closed-loop control system ensures accurate seed placement. Adding manure through our robot on the agricultural field is similar to the sowing mechanism. Funnels are being used to spray manure that spreads throughout the field.¹⁶ A control loop monitors the robot’s forward speed and triggers the servo gate every 20 cm along the row (pseudo-code included in Section 3.3.1). Gate timing was calibrated by bench tests to ensure one seed drop per actuation.

Weeding

Automation for weed control consists of two components: detection and actuation. Detection most commonly utilizes two-dimensional image processing to differentiate the plant from the soil by analyzing light reflectance, and to distinguish the crop from weeds by size differences and crop row patterns.^17,18 Actuation has two methods: one is to spray the weed and kill it with an herbicide; the other is mechanical, using a cultivator knife to remove the weed. The two-dimensional image processing techniques used for weed detection involve color-based segmentation and feature extraction. Color-based segmentation is an image-processing technique that separates objects depending on the color characteristics.^19,20

Feature Extraction is a process where individual features, such as color, shape, and texture, are extracted from raw data, allowing objects, like crops and weeds, to be separated and analyzed.²¹ Algorithms such as K-means clustering and Support Vector Machines (SVMs) are employed to classify plants as crops or weeds based on their color and shape characteristics. SVM is a supervised machine learning algorithm used for classification tasks to separate crops from weeds based on their features.²² Weed is detected using K-Means Clustering, an unsupervised machine- learning algorithm for data classification into clusters depending upon similarity. The DHT11’s data, processed by the Raspberry Pi, informs decisions about environmental control.

Images are processed using OpenCV pipelines, followed by classification with an SVM model trained on crop vs. weed datasets. Detected weeds are cross-checked against expected crop row geometry to minimize false positives. Classification accuracy is reported in Results as weed detection accuracy (%). The weed-detection algorithm employs an image-processing pipeline comprising color segmentation, feature extraction, and Support Vector Machine (SVM) classification. A dataset of 2,400 annotated images of maize crops and common weeds was used, split into 70 % training and 30 % testing sets. The trained model achieved a precision of 0.91 and a recall of 0.88, with a confusion matrix summarizing true and false classifications. Average inference latency on the Raspberry Pi 4B was 0.24 s per frame. All pseudocode for seed dispensing and classification logic has been summarized to support reproducibility, and the complete codebase will be made available upon request.

Harvesting

Harvesting is done using the parallel gripper, a robotic tool designed for grasping objects with two jaws that move parallel to one another.²³ The plants will be clamped and placed through a gripper on a conveyor belt. The parallel gripper is constructed from a PLA material using 3D printing technology.²⁴ The parallel gripper is actuated using an MG996R high-torque servo motor, which allows precise control of jaw movement. The opening and closing of the gripper are controlled via pulse width modulation (PWM) signals generated by the Raspberry Pi 4 B. To prevent damage to delicate crops, the gripper is programmed to operate within a restricted torque range, and the closing force is fine-tuned through servo calibration. The current system is a passive gripper with fixed jaw geometry, but it can be upgraded to an adaptive design in future iterations to accommodate varying crop shapes and sizes.^25,26 Simulation-based gripper force analysis was carried out in MSC Adams to verify clamping pressures. Prototype tests showed that the harvesting damage rate averaged 6% ± 1.2% across five trials.

Sensor Fusion

Our robot consists of sensor fusion techniques that combine the data from multiple sensors to improve the accuracy and reliability of the system. It can be calculated using a weighted average approach through which the sensor reads the environmental conditions, like soil moisture, temperature, and pH levels. This information is used to assist in decision-making mechanisms that guarantee that the robot can adapt to changing environments efficiently and effectively. Additionally, the system employs machine learning approaches, such as convolutional neural networks, to identify weeds and assess crop health. These advanced techniques allow the system to autonomously and efficiently complete the harvesting process. Each sensor (DHT11 for temperature and humidity, and pH sensor) contributes to the overall environmental condition assessment. For instance, pH readings are given higher weight due to their critical impact on crop selection, while temperature is moderately weighted, and humidity has the lowest weight due to known limitations in soil measurements.

Proposed System

In India, agriculture is dominated by farmers with typically less than an acre of land. Unlike large acres of farms in Western and Eastern countries, these small plots face soil compaction from heavy machinery, reducing crop yield. The proposed system is a lightweight robot designed to address the different challenges faced by small-scale farmers in India. Unlike the expensive and heavy machinery used in the fields that are often inaccessible, this robot is tailor-made to meet the needs of farmers with limited landholdings. Its chassis is constructed from lightweight material to prevent soil compaction. The robot can navigate effectively throughout farmland, and the system is designed to perform seed dispensing, weeding, and harvesting. The entire robot is much lighter than conventional farming equipment and easier to move and deploy in different field situations.

The Raspberry Pi 4B runs Python-based programs that process data received from sensors, control actuators, and execute decision-making algorithms (Figure 1). OpenCV is used for image processing tasks and the analysis of real-time data performed through sensors such as DHT11 and EZO pH sensors to optimize operations. Many existing agricultural robots specialize in a single process, which makes them very efficient for that particular task.^27–29 In comparison with other existing agricultural robots, the proposed system provides various benefits. This robot combines multiple functions in a single system, which decreases labor costs and vastly increases productivity by automating tasks such as sowing, weeding, and harvesting.

In addition, the proposed system’s design is customizable, which enables it to tackle various agricultural activities without the need for separate, specialized robots, unlike other robotic systems. It also lowers the complexity of deploying robotic solutions in small-scale farming. The 12 V Li-ion battery provides the main power supply, which is regulated through a buck converter to 5 V for the Raspberry Pi 4 B. An emergency stop relay ensures motor power can be cut off safely before reaching the TB6612FNG motor driver, which controls four DC motors. The Raspberry Pi generates PWM and direction signals for the motor driver as well as servo signals for three MG996R servos. Sensor data (DHT11, soil moisture, pH) is acquired through the MCP3008 ADC via SPI communication. The Raspberry Pi processes this data in real time to support autonomous decision-making, while also logging values for analysis. The integration of motors, servos, and sensors enables both mobility and task-specific operations. Safety features such as the relay prevent hardware damage during faults. Overall, the system architecture ensures efficient power distribution, reliable motor control, and accurate sensing for field deployment, while the Raspberry Pi Camera v2 connects directly through the CSI interface for image capture. Power, data, and control connections are shown in red, blue, and green, respectively.

Row-following vision algorithms are implemented for navigation, with optional GPS support in open fields. Safety features include an emergency stop switch that cuts motor power, watchdog timers, and actuator interlocks to prevent unintended tool operation. These measures ensure safe operation in unpredictable farm environments. The system combines sensor fusion, image processing, and actuation control to enable autonomous operation. Environmental data from soil and climate sensors are filtered and fused into a soil health index that guides servo-based interventions, while images from the Raspberry Pi Camera v2 are processed to detect crop rows for navigation. The Raspberry Pi generates PWM and direction signals for the motor driver and servos, ensuring accurate row following, with safety maintained through an emergency stop relay.

The weeding mechanism employs a mechanical blade actuated by an MG996R servo motor, operating at a depth of 1.5 cm for precise removal of weeds between crop rows. Navigation is governed by a vision-based PID controller that maintains alignment using real-time camera feedback on crop rows. Safety features include an emergency stop relay with a 92 ms response, watchdog timers to prevent control lock-ups, and manual override in case of sensor failure. These measures ensure reliable operation and prevent accidental crop damage during autonomous missions. The reproducibility of the proposed system has been enhanced through the inclusion of detailed technical data. The complete Bill of Materials (BOM) with component specifications, cost, and source links is presented for transparency. The robot’s total weight is 3.8 kg, with chassis dimensions of 45 × 35 × 25 cm fabricated using PLA and  aluminum supports. A wiring schematic showing sensor, motor, and controller interconnections is provided to support replication. Updated high-resolution CAD and mechanical drawings from TinkerCAD and MSC Adams illustrate the assembly layout and component placement.

Results And Discussion

The multipurpose agriculture robotic system was successfully designed and fabricated (Figures 2 and 3). The tasks implemented are weeding, harvesting, and sowing. This system uses a Raspberry Pi 4B as its microcontroller. Four 12V DC gear motors are used for mobility. MG996R servo motors are used for sowing and cleaning the land. The Raspberry Pi 4B (1.5 GHz quad-core processor, 4GB RAM) is used as the robot’s control unit. These eco-friendly approaches, instead of using any fossil fuel engine vehicles, reduce the impact on plants and soil and also eliminate unwanted weeds without harming crops, promoting sustainable agriculture. The sowing process is automated, with robots placing seeds at a specific predefined gap to optimize growth conditions; this ensures uniform plant distribution, reduces competition for nutrients, and maximizes yield.

The efficiency of using automation in the agricultural field largely helps to reduce labor costs, increase precision, and minimize soil compaction. Soil and climate sensors (temperature, humidity, pH, and moisture) provided input for environmental monitoring. The capacitive soil moisture sensor in the robot checks the soil’s moisture level. The automated sowing mechanism is operated through a servo motor and slider crank mechanism. These sowing cans are precisely placed at a specific distance and reset to the agricultural norms. This can be altered to any specific crop’s row and column distance between seeds. So, this helps the seeds to be sown uniformly in the field. Overall, we could infer from this proposed system that this could reduce seed wastage rather than throwing seeds randomly in a field without any particular distance between them.

The lightweight and mobile design of the robot enables smooth navigation across various terrains without affecting soil fertility. As an electronic system rather than a traditional large mechanical structure, it offers flexibility where the components can be easily plugged in or unplugged as needed. If only a specific motor or sensor is required, others can be removed effortlessly by disconnecting their terminals. This makes the system both user-friendly and convenient for farmers. This low power consumption of the root directly contributes to sustainability by reducing energy costs and carbon emissions. By far, the prototype results indicate that the multipurpose agricultural robot successfully integrates various automated processes to improve efficiency in farming operations compared to conventional farming methods; the robot aims to reduce labor dependency and optimize resource utilization, which enhances crop yield. Gripping and Sowing mechanics promote precision agriculture and farming techniques, leading to higher productivity and suitability. The aim is to optimize its power efficiency, scalability, and adaptability to different crop types and enhance its effectiveness in real-world agricultural applications.

The multipurpose agriculture robot presents a cost-effective, simpler robot mechanism that would be viable for small-scale farmers and would help them cut unwanted costs and increase agricultural production. By integrating automation in soil monitoring, sowing, weeding, and harvesting, the robot serves as a one-stop solution for various agriculture operations. Despite the promising results, the current prototype has certain limitations that need to be acknowledged. The system’s performance may be affected by uneven terrain, varying soil conditions, and extreme weather such as heavy rain or strong sunlight that can disrupt sensor readings. Crop height and dense canopy cover can also interfere with camera-based row detection, while sensor noise, particularly in low-cost soil probes, may reduce accuracy without frequent calibration.

For realistic field deployment, these constraints highlight the need for robust waterproofing, adaptive vision algorithms, and more reliable sensor modules. From an economic perspective, the lightweight and modular design makes the robot well-suited for smallholder farms, with an expected cost per hectare significantly lower than that of heavy machinery. Maintenance is simplified by the use of off-the-shelf components, and the system’s multifunctionality can improve return on investment by reducing labor costs and seed wastage across multiple operations. These considerations highlight the robot’s potential for affordable and sustainable adoption in real-world farming Additional field trials were conducted on maize and groundnut plots of 10 × 10 m to validate the robot under different crop geometries. Comparative analysis against manual operations and commercial robots (Naïo Oz, FarmDroid FD20) showed that MARS achieved similar or superior accuracy in sowing and weeding while maintaining low operational cost. Results are reported as mean ± SD with 95 % confidence intervals, confirming statistical significance (p < 0.05). These results strengthen the validity of the system and demonstrate its adaptability to multiple crop types and field conditions.

To provide quantitative validation, experiments were conducted on a 5 × 5 m maize plot under loamy soil conditions (pH 6.7–6.9, soil moisture 18–22%) at ambient temperatures of 28–32 °C and relative humidity of 55–65%. Each task – sowing, weeding, and harvesting was repeated in five independent trials. The system achieved a sowing spacing error of 2.3 ± 0.4 cm with a germination rate of 94% ± 2.1, while weed-detection accuracy averaged 91% ± 3.2 with a weeding success rate of 87% ± 2.7. Harvesting trials showed a crop damage rate of 6% ± 1.2. The robot maintained an average field coverage speed of 18 m²/hour with a battery autonomy of 3.5 hours. These values were reported as mean ± standard deviation across trials. When compared with manual sowing and weeding, the robotic system demonstrated superior precision and consistency, while maintaining crop damage within acceptable limits. Benchmarking against published agricultural robots further highlights that MARS achieves comparable or better performance in terms of precision sowing and weeding accuracy, despite being designed as a low-cost solution for smallholder farms.

A complete power budget analysis was performed to quantify energy usage. The Raspberry Pi consumed 3.5 W, four DC motors 28.8 W, servos 6 W, and sensors 1.5 W, totaling approximately 40 W during field operation. The 12 V Li-ion battery provided 3.5 hours of autonomy under a 70 % duty cycle. The total system cost was approximately ₹18,700, with operational expenses of ₹270 per hectare. Compared to manual labor, the robot achieved a payback period of less than one year and an estimated return on investment within eight months, demonstrating its practical economic viability for smallholder farmers. Future improvements, such as AI-driven decision-making and machine learning-based crop monitoring, could further revolutionize agriculture automation. Additionally, the robot can be made modular and customizable, allowing specific units such as sowing, weeding, or harvesting to be selectively activated or detached, enabling farmers to perform only the required operation efficiently without deploying the full system. The novelty of our system lies in its integration of multiple functions, such as sowing, weeding, harvesting, and soil monitoring into a single, lightweight, and low-cost robot. This is achieved through a combination of precise control algorithms and efficient mechanical designs.

Conclusion

This work presented the design and validation of a multipurpose agricultural robotic system that integrates sowing, weeding, harvesting, and soil monitoring into a single lightweight platform. Ground tests confirmed its ability to improve precision and reduce labor dependency, with consistent performance across multiple trials. By minimizing soil compaction and automating repetitive tasks, the system supports sustainable farming practices. In addition to technical feasibility, MARS demonstrates strong economic potential: its modular design reduces maintenance costs, improves return on investment, and offers a realistic pathway for adoption in smallholder farms. Future extensions with AI-driven crop monitoring and modular attachments can further enhance its scalability and impact, making MARS a practical step toward affordable precision agriculture.

References

1. Selvan SS, Wahid A, Patel A, Kumar V, Sahu P. Challenges in Indian agriculture. Agricultural Reviews. 2021;42(4):456–9.
2. Hamadani H, Rashid SM, Parrah JD, Khan AA, Dar KA, Ganie AA, Gazal A, Dar RA, Ali A. Traditional farming practices and its consequences. InMicrobiota and Biofertilizers, Vol 2: Ecofriendly Tools for Reclamation of Degraded Soil Environs 2021;(pp. 119–128). Cham: Springer International Publishing.
3. Raj R, Kumar S, Lal SP, Singh H, Pradhan J, Bhardwaj Y. A Brief Overview of Technologies in Automated Agriculture: Shaping the Farms of Tomorrow. International Journal of Environment and Climate Change. 2024;14(7):181–209.
4. Nawaz MF, Bourrie G, Trolard F. Soil compaction impact and modelling. A review. Agronomy for sustainable development. 2013;33(2):291–309.
5. Shaheb MR, Venkatesh R, Shearer SA. A review on the effect of soil compaction and its management for sustainable crop production. Journal of Biosystems Engineering. 2021;46(4):417–39.
6. R Shamshiri R, Weltzien C, Hameed IA, J Yule I, E Grift T, Balasundram SK, Pitonakova L, Ahmad D, Chowdhary G. Research and development in agricultural robotics: A perspective of digital farming.
7. Naik NS, Shete VV, Danve SR. Precision agriculture robot for seeding function. In2016 international conference on inventive computation technologies (ICICT) 2016;(Vol. 2, pp. 1–3). IEEE.
8. Hwang J, Shin C, Yoe H. Study on an agricultural environment monitoring server system using wireless sensor networks. Sensors. 2010;10(12):11189–211.
9. Yamasaki Y, Morie M, Noguchi N. Development of a high-accuracy autonomous sensing system for a field scouting robot. Computers and Electronics in Agriculture. 2022;193:106630.
10. Shafi U, Mumtaz R, García-Nieto J, Hassan SA, Zaidi SA, Iqbal N. Precision agriculture techniques and practices: From considerations to applications. Sensors. 2019;19(17):3796.
11. Utama YA, Widianto Y, Hari Y, Habiburrahman M. Design of weather monitoring sensors and soil humidity in agriculture using internet of things (IOT). Transactions on machine learning and artificial intelligence. 2019;7(1):10.
12. Muhammad Z, Hafez MA, Leh NA, Yusoff ZM, Abd Hamid S. Smart agriculture using internet of things with Raspberry Pi. In2020 10th IEEE International Conference on Control System, Computing and Engineering (ICCSCE) 2020;(pp. 85–90). IEEE.
13. Balamurugan C, Satheesh R. Development of raspberry pi and IoT based monitoring and controlling devices for agriculture. Journal of social, technological and environmental science. 2017;6(2):207–15.
14. Billa P, Venkatesh K, Sai JV, Lohith K, Kumar AS. Effective monitoring and protecting system for agriculture farming using IoT and raspberry pi. Materials Today: Proceedings. 2023;80:2917–20.
15. Azmi HN, Hajjaj SS, Gsangaya KR, Sultan MT, Mail MF, Hua LS. Design and fabrication of an agricultural robot for crop seeding. Materials Today: Proceedings. 2023;81:283–9.
16. Shivaprasad BS, Ravishankara MN, Shoba BN. Design and implementation of seeding and fertilizing agriculture robot. International Journal of Application or Innovation in Engineering & Management (IJAIEM). 2014;3(6):251–5.
17. De Rainville FM, Durand A, Fortin FA, Tanguy K, Maldague X, Panneton B, Simard MJ. Bayesian classification and unsupervised learning for isolating weeds in row crops. Pattern Analysis and Applications. 2014;17(2):401–14.
18. Jin X, Che J, Chen Y. Weed identification using deep learning and image processing in vegetable plantation. IEEE access. 2021;9:10940–50.
19. Gai J, Tang L, Steward BL. Automated crop plant detection based on the fusion of color and depth images for robotic weed control. Journal of Field Robotics. 2020;37(1):35–52.
20. You J, Liu W, Lee J. A DNN-based semantic segmentation for detecting weed and crop. Computers and Electronics in Agriculture. 2020;178:105750.
21. Tang J, Wang D, Zhang Z, He L, Xin J, Xu Y. Weed identification based on K-means feature learning combined with convolutional neural network. Computers and electronics in agriculture. 2017;135:63–70.
22. Ahmed F, Al-Mamun HA, Bari AH, Hossain E, Kwan P. Classification of crops and weeds from digital images: A support vector machine approach. Crop Protection. 2012;40:98–104.
23. Elfferich JF, Dodou D, Della Santina C. Soft robotic grippers for crop handling or harvesting: A review. Ieee Access. 2022;10:75428–43.
24. Hohimer CJ, Wang H, Bhusal S, Miller J, Mo C, Karkee M. Design and field evaluation of a robotic apple harvesting system with a 3D-printed soft-robotic end-effector. Transactions of the ASABE. 2019;62(2):405–14.
25. Aygün S, Güneş EO, Subaşı MA, Alkan S. Sensor fusion for IoT-based intelligent agriculture system. In2019 8th International Conference on Agro-Geoinformatics (Agro-Geoinformatics) 2019;(pp. 1–5). IEEE.
26. Torres AB, da Rocha AR, da Silva TLC, de Souza J N, Gondim RS. (2020). Multilevel data fusion for the internet of things in smart agriculture. Computers and Electronics in Agriculture, 171, 105309.
27. Shaik K, Prajwal E, Bonu M,Reddy BV. GPS based autonomous agricultural robot. In 2018 international conference on design innovations for 3cs compute communicate control (ICDI3C) 2018;(pp. 100-105). IEEE.
28. Amer G, Mudassir SM, Malik MA. Design and operation of Wi-Fi agribot integrated system. In2015 International Conference on Industrial Instrumentation and Control (ICIC) 2015;(pp. 207–212). IEEE.
29. Srivastava A, Vijay S, Negi A, Shrivastava P, Singh A. DTMF based intelligent farming robotic vehicle: An ease to farmers. In2014 International Conference on Embedded Systems (ICES) 2014;(pp. 206–210). IEEE.

Cite this article as:
Muthukaruppan V, Dhanush S A.M, Mirnalini S.S and Shruthi S. Development of a Multipurpose Agricultural Robotic System: An Experimental Study. Premier Journal of Science 2025;15:100205

Export to EndNote Export to Mendeley