AGROBOT: AN INNOVATIVE ARTIFICIAL INTELLIGENCE BASED SCARECROW ROBOT FOR CROP PROTECTION

Abstract

The main objective of our project is to design and develop an AI-based scarecrow robot to protect crops from invasive birds. A special feature of this robot is that it repels only invasive birds, not all birds. In this research work, a deep learning neural network model is developed to identify the 'farm disturbance bird'. Modified-MobileNetV2 was developed using a transfer learning approach to recognize ‘farm disturbance bird’ from images. Results of the proposed model indicate that it is able to recognize photo images with a good accuracy level and Real-time video images of 'farm disturbance bird' can be detected with good precision. An AI-based program is loaded into a controller with relays and a camera. Agrobot arms are activated when the AI program identifies the 'farm disturbance bird'. The proposed model is tested in a real-time environment with different distances and the results are recorded. It has been found that the robot is capable of identifying invasive birds up to 15 feet away. Ninety percent of the birds flew away from the place where the agrobot was active. A farmer can choose the 'farm disturbance bird' in the controller based on his or her experience. This agrobot assists in protecting the ecological system as it allows beneficial avian species to visit the farm. It also increases the farm's production rate by protecting it from birds that affect it.