Defining parameters and prototype manufacturing of an autonomous vehicle for orchard sprayer

In this study, an electric and autonomous agricultural vehicle, Vabot, was developed for pesticide spraying in orchards. It has four wheels, a GPS sensor, and can follow routes autonomously or be remotely controlled via Bluetooth connection. The vehicle's pesticide spraying mechanisms utilize artificial intelligence algorithms. The images captured through the camera are processed by three different models trained with the Yolov5 artificial intelligence algorithm, enabling the detection of leafy, leafless, and customized tree canopies for pesticide spraying. The spraying system includes a 7-bar pressure pump, a 0,8 mm TeeJet nozzle, a 25-liter tank, and relevant attachments. The vehicle can perform spraying in fixed or movable configurations using a crank-slider mechanism. Trials took place in apple orchards owned by Ankara University Faculty of Agriculture. The leafy model achieves a true positive rate of 97 % and a false negative rate of 3 %, while the leafless model has a true positive rate of 45 % and a false negative rate of 55 %. The trial orchard model attains a true positive rate of 77 % and a false negative rate of 23 %. The vehicle's field efficiency was determined as 41 %. The movable mechanism achieved a droplet diameter of 147-182 μm and a 60,43 % coverage rate, while the fixed mechanism achieved a droplet diameter of 125-200 μm and a 14,87 % coverage rate. Based on the conducted trials, it is predicted that the electric autonomous vehicle is suitable for pulverization applications in dwarf orchards.