Position Distribution Prediction Algorithm of Quadrotor Drones based on QRGRU

Konferenz: ISCTT 2021 - 6th International Conference on Information Science, Computer Technology and Transportation
26.11.2021 - 28.11.2021 in Xishuangbanna, China

Tagungsband: ISCTT 2021

Seiten: 6Sprache: EnglischTyp: PDF

Persönliche VDE-Mitglieder erhalten auf diesen Artikel 10% Rabatt

Lu, Jiahuan; Yang, Zhao; Zhang, Zhijie; Tang, Rong (College of Civil Aviation, Nanjing University of Aeronautics and Astronautics, Nanjing, China)

Real-time trajectory prediction is the guarantee for effective control of quadrotor drones. In this paper, based on the deep learning method, a position distribution prediction algorithm of quadrotor drones based on gated recurrent neural network quantile regression (QRGRU) is proposed. On the basis of GRU, QRGRU improves the loss function. The trajectory data is used for training and predicting, including nine variables: latitude, longitude, altitude, speed in the x direction, speed in the y direction, vertical speed, heading angle, pitch angle, and roll angle. QRGRU can master the inherent laws of quadrotor UAV flight by training and learning the historical trajectory of quadrotor UAV. The results show that QRGRU effectively predicts the position distribution of quadrotor UAV. Compared with other methods, QRGRU improves the prediction accuracy, model reliability and sensitivity, shortens the prediction time, and has a good effect on multi-step prediction. The model provides conditions and more time for taking control measures in time.