Optimizing Pesticide Application Via Drone Navigation: A Reinforcement Learning Framework
Main Article Content
Abstract
In modern-day agriculture, efficient and targeted pesticide application is paramount for protection of crops, environment sustainability and safeguarding the farmers’ health. Reinforcement learning (RL) algorithms have shown promise in optimizing such tasks by enabling autonomous decision-making. This paper delves into the development of an environment that makes use of Reinforcement Learning techniques, specifically reward shaping, to train autonomous drones for efficient crop spraying. This research aims to investigate whether a Reinforcement Learning-based environment can be designed to effectively guide drones in navigating and spraying crops, effectively answering the critical question. By doing reward shaping, a reward function was discerned that optimizes for different and basic parameters crucial to farmers.