中文 |

Researchers Proposed CNN-based Estimation of Aircraft Landing Gear Angles

Author: LI Fuyang |

The neural network has great application value in aviation security. Using the neural network method to monitor the status of aircraft landing gear can improve the intelligence, stability and objectivity of airport operation.
In a study published in Sensors, a research group led by Prof. WU Zhiguo from the Changchun Institute of Optics, Fine Mechanics and Physics (CIOMP) of the Chinese Academy of Sciences (CAS) proposed a multi-step CNN-Based estimation of aircraft landing gear angles.
The algorithm includes target normalization module, pixel voting module and angle calculation module. Input a single RGB image, the algorithm first detects the key points of the aircraft and then calculates the angle. The plane is divided into two parts to predict key points. The key points of fuselage are detected by target normalization module and the key points of landing gear by pixel voting module. Then angle calculation module outputs landing gear angle without depth information. In addition, aircraft simulation data sets are made based on the aircraft CAD model.
The target detection and normalization module normalized the scale of the aircraft target to ensure that the aircraft has a higher resolution and improve the accuracy of Angle detection. The Angle calculation module overcomes the problem of lack of depth information and avoids the regression of redundant spatial parameters, thus improving the Angle measurement accuracy. The loss function is optimized based on pixel distance, the learning proportion of main pixel is enhanced, and the pixel error of key points of the fuselage is effectively reduced. 
The simulation aircraft data set with variable landing gear angle are made based on the aircraft CAD model for landing gear angle estimation. The error of the Angle predicted value of the proposed algorithm is 4.6 degrees, and the running speed of the algorithm is more than 16 fps.
The algorithm can detect the landing gear angles, adapt to the trend of intelligent and unmanned airport, and play a positive role in improving airport operation efficiency and ensuring flight safety.
Contact

WU Zhiguo

Changchun Institute of Optics, Fine Mechanics and Physics

E-mail:




       Copyright @ 吉ICP备06002510号 2007 CIOMP130033