Abstract :

Autonomous drones are increasingly used in applications such as surveillance, delivery, disaster management, and agriculture. A key challenge in drone autonomy is efficient and safe path planning in dynamic and complex environments. Traditional path planning algorithms rely on predefined maps and handcrafted rules, which limit adaptability. This paper presents an autonomous drone path planning approach using deep learning techniques. The proposed system enables drones to learn optimal navigation strategies directly from sensor and environment data. Deep neural networks are used to predict collision-free paths while minimizing travel distance and time. The model adapts to dynamic obstacles and uncertain environments. Simulation-based training improves robustness and generalization. Experimental results show improved navigation efficiency compared to conventional methods. The proposed approach enhances autonomy and decision-making capabilities of drones. This work demonstrates the effect