Autonomous Trajectory Planning For Unmanned Aerial Vehicle

Abstract

Unmanned aerial vehicles (UAVs) have a significant problem in autonomous navigation in new or unpredictable situations.To address this issue, the system produces a path between two points, and the drone is commanded to follow this path based on its location.In this study, we provide a novel framework for autonomous UAV route planning based on deep reinforcement learning.The goal is to approach moving or stationary targets using a self-trained drone (UAV) as a mobile aerial unit in a 3-D urban setting.UAVs, particularly rotary-wing aerial robots such as quadcopters, offer a high level of mobility, making them appropriate for a wide range of activities and applications.To provide a safe autonomous flight with limited mission time or battery life, the most efficient path must be determined.The quadrotor UAV in our trials continually monitors its position and battery level and modifies its course accordingly.We simulate the behaviour of autonomous UAVs in several conditions, including obstacle-free and urban environments.Our findings show that the UAV is capable of choosing clever paths to its objective.