NARX-based Trajectory Planning for Obstacle Avoidance in Precision Agricultural Drone Operations

Precision agricultural drones require advanced obstacle avoidance to navigate dynamic environments efficiently. Traditional methods lack adaptability to real-time changes. We propose a novel NARX model-based approach, leveraging real-time sensor data and a vertically positioned radar for dynamic trajectory planning. The pilot controls pitch, yaw, and roll, while the model manages throttle using radar data to gauge the height and distance of upcoming obstacles. Our comprehensive framework, tested in Gazebo simulations, demonstrates high obstacle avoidance accuracy and computational efficiency. This advancement enhances the safety and efficacy of drone operations in precision farming.