Plant Disease Detection and Automated Spraying Robot: A CNN and IoT-Based Approach for Precision Agriculture

Abstract

Agriculture is a vital sector that directly influences food security and economic stability, yet crop productivity is significantly affected by plant diseases. Traditional disease detection methods rely on manual observation, which is time-consuming, subjective, and often inaccurate, especially over large agricultural fields. To overcome these limitations, this paper presents a Plant Disease Detection and Automated Spraying Robot that integrates image processing, deep learning, and robotic automation to enable precision agriculture. The system uses a smart camera to capture real-time images of plant leaves, which are analysed using a Convolutional Neural Network (CNN) trained to classify healthy and diseased leaves such as rust, powdery mildew, and viral infections. Based on the detection results, an ESP32-controlled mechanism automatically activates targeted pesticide spraying, ensuring accurate and timely treatment of infected plants. Obstacle detection and navigation are managed using ultrasonic sensors and motor drivers, enabling autonomous movement across the field. Cloud integration using Firebase allows real-time monitoring through a web dashboard that displays disease type and system status. Experimental results show an average detection confidence of 76-85% and a spray-activation accuracy of 100% when disease confidence exceeds the 60% threshold, with a total system response time of under 5 seconds. The proposed system reduces manual labour, minimizes pesticide wastage, limits farmer exposure to harmful chemicals, and improves crop yield.

Introduction

Agriculture plays a vital role in human survival and food security, but plant diseases, climate change, pests, and soil degradation significantly reduce crop yield and quality. Traditional plant disease detection through manual observation is time-consuming, subjective, and prone to errors. Therefore, automated systems are needed to detect diseases quickly and improve crop management.

This paper proposes a Plant Disease Detection and Automated Spraying Robot that uses a Convolutional Neural Network (CNN) to identify plant diseases from real-time images and automatically spray pesticides on infected plants. The system aims to reduce chemical wastage, labour costs, and farmers’ exposure to harmful pesticides while improving treatment accuracy and crop productivity.

The literature survey shows that existing CNN-based approaches achieve high disease classification accuracy, but many lack autonomous navigation, automatic spraying, or complete integration between disease detection and treatment. A field survey conducted in the Rattihalli agricultural region, focusing on tomato, cotton, and brinjal crops, also highlights the need for an efficient, field-deployable solution.

The proposed system collects images from the PlantVillage dataset and a Maizic smart camera. The images are preprocessed, resized to 224 × 224 pixels, and used to train a CNN model implemented with TensorFlow and Keras. Firebase and a Flask-based REST API support cloud communication and disease classification. The ESP32 microcontroller receives the classification results and controls the pesticide-spraying mechanism.

The robot uses an ultrasonic sensor to detect obstacles and locate plants within a distance of 30 cm. It captures plant images and sends them to the cloud for analysis. If the plant is healthy, the robot continues scanning. If a disease such as rust, powdery mildew, or leaf virus is detected, the corresponding pesticide pump is activated automatically.

The main hardware components include the ESP32, Raspberry Pi Pico, Maizic smart camera, ultrasonic sensor, L298N motor driver, relay module, and DC geared motors.

Conclusion

This paper presented a Plant Disease Detection and Automated Spraying Robot that integrates image processing, CNN-based deep learning, embedded control, and cloud communication to automate disease identification and targeted pesticide spraying. Experimental evaluation demonstrated reliable detection performance (76-85% average confidence), 100% spray-activation accuracy above threshold, and a total system response time below 5 seconds, while reducing manual labour, chemical wastage, and farmer exposure to harmful chemicals. Future work will focus on fine-tuning pre-trained models with additional local crop imagery to improve accuracy, extending multi-crop and multi-disease support using the same hardware platform, incorporating GPS-based autonomous navigation, and enabling fully autonomous spray-decision-making based on confidence levels without manual oversight — advancing the system toward a complete, scalable smart-farming solution.

References

[1] A. O. Santhosh, A. Sagar, M. Govindaswamy, S. Anees, and M. K. Monica, “Agni-A: Plant disease detection robot using deep learning,” Int. J. Eng. Res. Technol. (IJERT), vol. 12, no. 5, 2023. [2] A. Kamilaris and F. X. Prenafeta-Boldú, “Deep learning in agriculture: A survey,” Comput. Electron. Agric., vol. 147, pp. 70-90, 2018. [3] S. Sladojevic, M. Arsenovic, A. Anderla, D. Culibrk, and D. Stefanovic, “Deep neural networks based recognition of plant diseases by leaf image classification,” Comput. Intell. Neurosci., vol. 2016, 2016. [4] P. S. Priyanka, R. Meghana, and R. Sahana, “Automatic crop disease detection using CNN and image processing,” Int. J. Adv. Res. Comput. Commun. Eng., vol. 11, no. 4, 2022. [5] V. Ramesh and K. Gokul, “Design of an autonomous agricultural spraying robot using sensors,” Int. J. Innov. Res. Sci. Eng. Technol., vol. 8, no. 6, 2019.

Copyright

Copyright © 2026 Vaishnavi B N, Tejaswini S, Chandana T, Venkatesh R Shetter, Vanishree H V. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.