This paper proposes a new model-free UAV flight trajectory prediction method based on maneuver recognition, which avoids the complex modeling and computational problems of model prediction. Firstly, the recurrent neural network is trained to recognize maneuver with the mean square error as loss function. After obtaining a good maneuver recognition rate, the flight trajectory matching algorithm is used to match the historical flight trajectory with the template trajectory, through trajectory cutting and splicing to realize the prediction of the UAV flight trajectory. The effectiveness of the method proposed in this paper is verified by simulation experiments.
A Trajectory Prediction Method Based on Maneuver Recognition
Lect. Notes Electrical Eng.
International Conference on Autonomous Unmanned Systems ; 2022 ; Xi'an, China September 23, 2022 - September 25, 2022
Proceedings of 2022 International Conference on Autonomous Unmanned Systems (ICAUS 2022) ; Kapitel : 265 ; 2864-2873
10.03.2023
10 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
British Library Conference Proceedings | 2014
|EPOXI Trajectory and Maneuver Analyses
NTRS | 2011
|MULTI-MODAL TRAJECTORY PREDICTION OF SURROUNDING VEHICLES WITH MANEUVER BASED LSTMS
British Library Conference Proceedings | 2018
|