The purpose of this study is to explore the application of Deep learning (DL) in autonomous flight path planning of UA V, and improve the autonomous navigation and flight ability of UA V in complex environment. In terms of methods, this article first constructs a DL model to extract environmental information from sensor data carried by UA V. The model adopts the structure of Convolutional neural network (CNN), and identifies obstacles and safe flight areas through training and learning. Then, combined with the optimization ability of traditional path planning algorithm, using the perceptual information provided by DL model, an optimized path from the start point to the end point is planned for UA V. Experimental results show that the proposed algorithm can effectively plan a safe and efficient flight path for UA V in various complex environments. Compared with other path planning methods, the proposed algorithm has obvious advantages in path length, flight time and obstacle avoidance ability. In addition, the success rate and security of the algorithm have also been verified, showing the great potential of DL in UA V path planning.
Unmanned Aerial Vehicle Autonomous Flight Path Planning Algorithm Based on Deep Learning
29.05.2024
1296729 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
DOAJ | 2023
|Unmanned Aerial Vehicles Path Planning Based on Deep Reinforcement Learning
Springer Verlag | 2019
|