Smart Pesticide Sprayering Robot for Fertigation Farm

Abstract

This Smart Pesticide Spraying Robot is designed to improve pesticide use in fertigation systems by supporting more accurate and eco-friendly farming practices. It uses embedded AI on an ESP32-CAM to detect crops grown in polybags and spray them as needed. The object detection model was trained using Edge Impulse with three types of datasets: black polybags (V1), white polybags (V2), and a combined version (V3), labelled through Roboflow. The model trained on V3 achieved 93.06% accuracy and 95% precision in polybag detection. Once a polybag is detected, the robot activates a spray pump using a relay and controls its movement using two DC motors through an MDD3A driver, guided by the object’s location in the video frame. The system runs fully on-device without needing cloud access, thanks to Edge Impulse’s optimized inference engine. However, during testing, some issues were observed. The robot sometimes failed to respond or sprayed incorrectly, especially when nearing a target. This was likely due to the ESP32-CAM’s limited processing power and the absence of advanced motor control systems like PID. Despite these limitations, the project proves that low-cost, AI-powered robots can be used for smart farming. For future improvements, it is recommended to use a more powerful processor like a Raspberry Pi, improve motion control, expand the training dataset, and add more sensors for better navigation and accuracy.