In this, we presented a comprehensive methodology for an autonomous drone delivery system, focusing on efficient navigation, real-time obstacle avoidance, and precise control. The proposed system integrates Simultaneous Localization and Mapping (SLAM) for accurate localization within the environment and mapping of the environment, combined with the A* algorithm for optimal path planning. To enhance the system's capability to navigate dynamic environments, we utilize a deep learning-based model for real-time obstacle detection for collision avoidance. Furthermore, Model Predictive Control (MPC) is employed to fine-tune the drone's flight path, ensuring smooth and stable movement while adapting to environmental uncertainties. This multi-layered approach enables the drone to autonomously map its surroundings, compute the most efficient route, and safely avoid obstacles during delivery operations. The proposed method offers a robust solution for autonomous drone navigation in complex and unpredictable environments, making it highly suitable for drone delivery applications in various real-world scenarios.
Dronav: Deep Learning Based Autonomous Drone Navigation System
06.11.2024
520014 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch