The application of Deep Reinforcement Learning (DRL) to enhance the autonomous navigation capabilities of drones in environments that are complex and congested. The complications involved with guiding drones across restricted places, overcoming obstructions, and navigating among clutter are addressed in this article. Through the utilization of DRL techniques, the framework that has been proposed gives drones the ability to learn and alter their navigation strategies on their own through the process of trial and error, hence optimization of real-time decision-making. The incorporation of deep neural networks for the processing of sensory data makes it possible for drones to comprehend their environment, which in turn makes it easier for them to make educated judgments with regard to safe navigation. In order to demonstrate the potential of DRL in boosting the autonomy and robustness of drone navigation systems in demanding environments, the effectiveness of the technique is evaluated using simulations and experiments conducted in the real world. The findings make a contribution to the development of autonomous drone technology, which has implications for a variety of applications including environmental monitoring, search and rescue operations, and surveillance.
Deep Reinforcement Learning for Autonomous Drone Navigation in Cluttered Environments
2024-04-09
3784200 byte
Conference paper
Electronic Resource
English
Robotic Planetary Exploration : Autonomous Navigation in Cluttered Unknown Environments
British Library Conference Proceedings | 2012
|